29 releases

更新日誌

Every released version, newest first. This is the same file FluidEQ shows you in its What's new dialog the first time you run a new version — not a marketing summary of it.

Latest: v2.2.0 FluidEQ 首頁

Latest signal v2.2.0 29 releases

2.2.0

Latest

FluidEQ 2.2 speaks Korean and Traditional Chinese, tells you what is new in a notice that never gets in your way, and puts a scene behind your Karaoke lyrics with its own transparency and blur. Your scenes says plainly what is yours, what is waiting for review and what failed to load, and an empty Visualizers column now points you to the gallery and the Studio.

Languages

  • 한국어 and 繁體中文. FluidEQ now has twelve languages. Traditional Chinese is picked for Taiwan, Hong Kong and Macau, Simplified for mainland China and Singapore, and a language you chose yourself is kept. The installer speaks both too.

Announcements

  • News from FluidEQ in a floating notice. An announcement shows its headline and one line of its text over the window, without moving anything or blocking what you are doing, and is remembered once read. Some announcements are for signed-in accounts only.

Visualizers

  • Your scenes tells you what is happening. Scenes you own, scenes you published, scenes waiting for review and scenes that could not load are each shown as what they are, instead of failing silently.
  • An empty column points somewhere. With no visualizer scenes yet, the look picker offers Explore scenes and Create your own.
  • Your own scene lights your keyboard and mouse at once. A scene you made counts for Dynamic lighting while it waits for review, as it already played on the graph.
  • More scenes, one press away. Under your own scenes, the look picker's Visualizers column offers the gallery, and says how many scenes went up there since you last opened it.
  • Scene update notices show once per version and stay dismissed when you switch views or reopen the app.

Karaoke

  • A scene behind the lyrics, with its own look. Karaoke can show a visualizer scene or the graph behind the words, with its own transparency and blur, apart from the graph's own, and the scene follows Karaoke's mute.
  • Lyrics on an open stage. The phrase being sung is larger and sits in the middle of the stage; a long one wraps onto balanced rows instead of shrinking, with the highlight running across them on the song's own timing, and a soft veil under the words keeps them readable over a bright scene.
  • Exclusive cover and graph backgrounds for the Karaoke stage, kept apart from the graph's controls. Either one now reaches the edges of the page, with no frame of empty space around it.
  • Stage lights that follow the music. With no cover and no scene behind the words, two washes and two beams light the stage: the washes swell with the bass, the beams open with the level and the treble, and every beat flashes them.
  • Playlist folders fold. A folder in the Karaoke playlist opens and closes from its heading, which counts its songs.

Starting with Windows

  • FluidEQ starts in the tray. With Start with Windows on, it no longer opens its window at every sign-in: it starts in the notification area, shaping your sound from the first second, and its tray icon opens the window when you want it. Opening FluidEQ yourself still opens the window. A startup entry from an earlier version moves over by itself, and stays off if you switched it off in Windows' Startup apps.

Share Audio and Forum

  • Share Audio holds steady under load. Capture and the network link run in their own service, so a busy window no longer stalls the sound.
  • Fluid, for music over Wi-Fi. A switch beside Share plays every link about a third of a second later, and a little more after a dropout, so a stalling network no longer cuts the music. Off, Share Audio sounds exactly as before. Both computers need FluidEQ 2.2.
  • Forum issue threads update at once, and can be answered right in the thread once FluidEQ's GitHub permission allows it.

What's New

  • A guided tour of the whole app. What's New opens on one video that goes over every part of FluidEQ, with a voice saying how to use each one. It is voiced in English for now, with captions in all twelve languages, and it takes the place of the single-feature videos.
  • Slides for Karaoke backgrounds, starting in the tray and the two new languages, each with the real control. The tour reopens only for a new feature release, never for a fix release of the same one.

FluidEQ 2.1 gives every output a sound of its own. Put your headphones on and they sound the way you tuned them; switch to the speakers and their preset, DSP rack and Room come back with them. Any band can now be tuned for each ear on its own, the DSP rack gains a De-esser that turns down only a voice's sharp S, a second output plays straight from the FluidEQ Engine, Share Audio links a whole group of computers both ways, songs can remember how you like them, and the graph shows every band's numbers right on the curve. And the visualizers are yours for good: a scene costs one credit and stays yours, with no subscription, and the Studio is open to every account.

This release brings FluidEQ Engine 1.24. FluidEQ offers the update after it starts, with one Windows prompt; until then each ear plays both ears' bands, the De-esser is not heard and the leveling's output loudness is not shown, and everything else plays as before.

Outputs

  • Every output keeps its own sound. The EQ, the preset, the Tone, the cuts, the DSP rack, the Room and the leveling are saved with each output and come back the moment you switch to it — headphones and speakers no longer share one preset.
  • A second output plays from the FluidEQ Engine itself, with its own sound, and the main output never waits for it. Edit sound beside the second output edits it without switching the main output; Done or Edit main takes you back.
  • The second output keeps time by itself, and every delay is shown in milliseconds, the network's included when the sound comes from another computer.
  • Share Audio links a whole group, both ways. Link a computer once, from either side, and it joins the group: every computer sends its own sound straight to the others and plays theirs, with Receive and Send on their own switches for each one. Your mix holds this computer's volume and a volume of its own for every computer you hear. Add computer opens the pairing whenever you need it, computers that are away stay listed, and the group finds itself again by itself after a restart, a sleep, a network drop or a switch of audio engine. Each computer can be paused and resumed, the sound arriving shows the rate it comes in at, and a new connection always starts playing. The bar shows only the computers whose sound reaches you, and This computer says when sound is playing that no player names, such as a game.
  • Bluetooth headphones show their battery beside the output, and beside the second output.

The equaliser

  • My curves: your own list beside OPRA's. EQ Presets has a My curves tab next to the OPRA library. Keep any curve there under a name of your choosing: pasted from Squiglink or AutoEQ, read from a .txt file, taken from your EQ as it plays, or copied from the OPRA correction playing. The OPRA tab and the Squiglink import keep their curve with one press of Save to My curves, under the name it already has, and then say it is in the list. Squiglink has a tab of its own, right after the OPRA library, so a growing list never pushes it down. A line that cannot be read is named, never dropped without a word. Each curve shows its shape, where it came from and which ears it is for, and plays with one press as the headphone correction or goes into your EQ bands (that one asks for a second press, because it replaces them). The curve playing now is marked. Search, sort by name or by date, rename, edit its bands, duplicate, copy as text, delete — each from a menu with its own icon, and deleting asks first, naming the curve. Export list writes them all to one file that another FluidEQ imports whole, or to a folder of .txt files for Equalizer APO and AutoEQ tools. Import list shows every curve before adding any, names whatever it leaves out, and never overwrites a name you already have without asking: keep both, replace or skip. Your curves are kept in a file of their own in FluidEQ's data folder.
  • The OPRA list draws every measurement's curve beside its name, so the measurements of one headphone are told apart by eye, and Save to My curves stands beside Apply, ready before anything is applied. The curve playing is marked Applied in every tab, whichever tab applied it, and the correction's chip calls it by its name in My curves. The headphone picked links to its page on fluideq.com, where every correction for it is drawn and credited.
  • Switch corrections in one press. The EQ mode card's Corrections tab has a Correction picker: your curves first, then the OPRA curves for the headphone playing, with arrows either side to step through them, so comparing two corrections is one press. Each output keeps the one it plays.
  • Tune each ear on its own. Right-click a band and choose Left + right: it becomes two bands at the same frequency, one for each ear, that you move separately. Both ears joins them again in their middle, and Reset band and Clear EQ join them too. The graph draws the left ear solid and the right dashed. The same menu does it in the DSP rack's EQ while it applies to Stereo.
  • Delete band is in the band menu, in the main EQ and the DSP rack, for one band or a selection. It asks for a second press before it deletes. Deleting one ear's half of a pair puts the other half back on both ears.
  • Import into one ear. Squiglink's Apply as EQ, and the DSP EQ's Import, can fill Both ears, the Left ear or the Right ear. One ear replaces only that ear's bands and leaves the other ear sounding as it did.
  • Picking a built-in layout starts it fresh: 10, 15 or 31 bands at that layout's own frequencies and Q, flat and on both ears. It used to bring back what you last had on that layout.
  • Band labels on the graph. Each band can show its frequency, Q and gain over its point (View → Everything, or Ctrl+W). A label sits above or below its point, never beside it, keeps clear of the style dots and the analyser's key, stays put while you drag its band, and goes back to its own place once there is room. Resetting the EQ lays them out fresh.
  • EQ mode sits beside the graph. Its settings are the first card of the side panel on the EQ page, Your EQ and Corrections one at a time, so they no longer cover the curves they change. The EQ mode button brings the card into view, opening the panel if it was closed; the pin on the card puts the settings back in a menu.
  • Corrections are smoothed at 1/12 octave until you choose otherwise, and Reset in EQ mode goes back there. Off is still one press away.
  • The curve follows a slider at once. Moving a band's slider redraws the curve in the same frame, as dragging the curve already moved the slider.
  • The Tone knobs and a band's controls change places smoothly. Selecting a band fades the Tone row out and brings the band's controls in from left to right; clicking off brings the Tone knobs back the same way, and moving from one band to another does too. Each change plays once, dragging a band moves nothing, and Animations off makes it instant.
  • The handles and knobs work from the keyboard: the arrow keys step a fader, a knob or a point's gain, and a screen reader names each point by its band and frequency. While the EQ is switched off its controls are dimmed and cannot be changed by accident.
  • A new layout starts every band at the same Q, from how many bands it has and how wide it spans, and a band you add takes the Q of the ones already there. Your bands' Band Q no longer reshapes a preset or a correction.
  • The Smart EQ toolbar was tidied, and the Applied headphone button stays readable under the mouse.
  • Expanded grows the graph in its own card. On the Bands page the page folds away under it and the head stays where it was, rounded and floating on the Backdrop as at the normal size. Full screen, and Expanded over Media, the Library and Karaoke, still lay the graph over the whole column.
  • Game presets have a switch of their own at the head of their page. Off, no game changes your sound and FluidEQ stops watching which program is in front; a game playing when you switch it off gives your sound back at once. Your list stays, and Reset settings turns it on again.

The DSP rack

  • De-esser. A new stage, after the restorations and before the Exciter, that turns down the sharp S, T and SH of a voice only while they sound and leaves cymbals and hi-hats alone. It reduces the S's own frequencies (Spectral), everything above it (Treble) or the whole voice; it listens in the centre of the stereo picture or to all of it; and Isolate plays only what it takes away. A four-second strip beside the treble shows each S and the cut over it, with the cymbals it let through in grey. Seven profiles to start from, for music and for speech; Speech, Podcast and Audiobook switch it on. Switching it on or off never moves the sound in time. Beside the strip, the treble as it is now shows how far the S stands over the voice, and how much more a cymbal would need before it is touched.
  • Exciter oversampling: Off, Standard or High, and Off to start. Off is the lightest; Standard and High run the Exciter faster so nothing folds back off the top, High with the sharpest filter. Picking any factory preset turns oversampling off, in the Exciter and in the EQ alike; turning it on stays your choice. Oversampling now takes far less work than before, which is what made the sound cut out on some computers with High on.
  • The pop-style genre chains add their brightness with the Exciter.
  • Live leveling follows your target at once. Changing its loudness target or its peak ceiling glides the level to the new one in the middle of a song, without forgetting what it learned of the song. The panel shows the loudness going in and the loudness coming out, each over the last three seconds.
  • Save on every filter. Denoise, De-esser, Exciter, Bass Forge, Bass Punch, Dimension, Maximizer and Master each keep profiles of your own, listed first under Yours, saved and deleted the way the EQ's are.
  • The Exciter's bars no longer go dark on Mid only and Sides only, and its Timing no longer moves part of a sound from one side to the other; switching between its modes no longer jumps in level.
  • The starred presets are listed in the same order as the full list, with each Room copy right after its chain.

Songs

  • Remember each song's sound. Switched on, a preset, Tone or EQ you set during a song comes back whenever that song plays here, with a notice that says so and an Undo. It starts switched off. One player at a time starts switched on.
  • A playlist stays in Up next. Playing a playlist, from the Library or the Compact player, keeps that playlist in Up next whatever you browse next; Add to up next puts songs at the bottom, each once, even when the same playlist is added twice. With Keep playing on, More like this appears only when the last song is playing, as its own part below the list. Music dropped on the Compact player's Up next goes in where you let go.
  • A Library scan can be put out of the way. Its notice has a × that hides it while the scan carries on, and a small Stop; the Rescan chip says how far the scan has got and brings the notice back.

The Compact player

  • One Compact player, the same in every window colour. It is one column: the visualizer at the top with the song over it, the transport, the Sound with its Tone knobs or your EQ's bands, and Up next and Playlists below, each folding away when you want the room. Drop music anywhere on it and it goes on the end of Up next. With a scene from Visualizers it is always the Backdrop, and the scene follows Brightness as the full app does. It has one width; its height is yours, from its top edge or its bottom one, and nothing in it scrolls but the list. It replaces the classic amp and the Stage.
  • It opens and folds with a motion of its own. From the full app or from the strip folded to one line, the column opens down out of its header, and folding to one line plays it back up into it. The full app switches without motion, to stay light, and Animations off makes both instant. Coming back from it puts the full app where it was, at its own size.
  • Folded to one line, it has room to breathe. The strip is a little taller, with even margins, a soft play key, quieter window buttons and a thin line for the song's position that never glows; its knob shows when you point at it. A standard visualizer moves faintly behind the song, as a scene already does.

Visualizers

  • Any colour for the window with a standard visualizer. A Tint slider beside Brightness, in Window colours, the main menu and the Compact player's menu, turns the whole window to the colour you pick, the graph and the faders with it. Double-click it to go back to the usual colour; Reset all settings does the same.

  • Rainbow mode is now Colourful mode, and it plays in the window's own colours — the Tint you picked or a scene's — instead of a fixed rainbow. Its share card is new too.

  • Tint app icon. A switch under Colourful mode, in the main menu, Window colours and the Compact player's menu, draws FluidEQ's taskbar button, its tray icon and the icon in the tray's menu in the window's colours, and follows them as they change. It starts off.

  • The name in the title bar stands alone, on one line in one sweep of colour, with the version coming in last.

  • What a scene's time of day follows is one choice of three, Off, Brightness or Clock, right under its Daylight slider, in Window colours and in the Compact player's menu. Daylight always reads as a whole percent.

  • Scenes are yours for good, one credit each. Open a scene in Visualizers → Gallery, watch it for 10 seconds, and Get keeps it in your looks for good — on the graph, your desktop and your lights, on every computer you sign in to, with every update its maker publishes. Credits come in packs paid once on Stripe in your browser — 5 for $4.99, 12 for $9.99, 30 for $19.99 — and nothing renews; a confirmed account gets 3 credits free to start. Credits you have not spent are refunded if you ask within 14 days. Checkout opens in your language, and credits paid in the browser reach FluidEQ by themselves, through Back to FluidEQ on the page after paying.

  • No more subscription. The Plus membership is no longer sold, and the tab it opened is now Visualizers. A membership already running keeps working until FluidEQ moves its members to credits; then every scene a member added stays theirs, and every account that paid for it gets at least 5 credits.

  • FluidEQ can make some of its own scenes free for everyone: while one is free it plays for every account signed in.

  • The leaderboard is gone.

  • Tune the wave where you put a visualizer on the desktop. The Desktop background dialog has Wave height and Wave position, drawn at once on the monitor that will play it, and opens on the wave a monitor already plays it with.

  • Standard looks have their own graphics settings, apart from the scenes: full resolution and plain scaling to start with.

  • Tap a 3D world's birds or fish and they fly off the way they face, larger near the bottom of the screen and smaller towards the top, as they would be nearer or farther away. They used to spin like petals and fly on their backs.

  • Galaxy, Nebula Heart and Nebula draw many times faster on integrated graphics (Galaxy from 16 ms to about 1 ms a frame on Intel UHD), and Supernova no longer loses its picture there. Glass in the 3D worlds costs less too (Prism Bloom from 20.6 to 16.1 ms a frame on Intel UHD at 1080p).

  • New sets of scenes are announced in the app, once each, with the set's scenes framed and Try them now to play the first on the graph.

  • Hearts on members' scenes. Like a scene in the gallery and its maker sees it; FluidEQ's own scenes carry no counts.

The Studio

  • The Studio is free for every account. Make scenes with your AI, export them and publish them to the gallery, with no membership. A scene you publish is read by FluidEQ's maker first, and plays free for you everywhere once it is approved; each time an account gets it with a paid credit, you earn half a credit, and every two make a credit to get other members' scenes or FluidEQ's. The account panel says when half a credit is waiting.

  • Ask for a change and watch it on the stage. The Studio has a chat beside its stage: say what should change — how a scene looks, how it moves with the music, its colours or its light — and your own AI edits the scene and plays each version as it goes. It runs on Claude Code installed on your computer, signed in with your own plan; FluidEQ never sees the sign-in.

    • Every reply keeps the version it made, with Bring back; each scene keeps its past chats, and chats Claude Code kept for the project's folder can be opened too.
    • Attach the stage sends a picture of the scene with your message, and you can draw on it — arrows, words, marks — to show what you mean.
    • Messages you send while it works are read at its next step, not after the reply, and Stop keeps them going. Signing in from a failed reply sends it again, and every picture in the chat opens full size.
    • Pick the model and how hard it thinks; a ring shows how much of the conversation's room is used, Compact makes more, and the accounts card shows how much of your 5-hour and weekly limits is left.
  • The message you are writing is kept. Each scene keeps its unsent chat message, and the stage picture attached with your drawings on it, through a restart, until you send it.

  • The Studio says which versions you have: the bench's, the copy in your looks (and whether it is behind), and the gallery's.

  • While the Studio is open, the scenes' time of day follows its Brightness, so a scene can be checked at night and in daylight before it is published.

  • The Studio's frame rate is the picture's. It counts the frames the scene actually puts on screen, so a busy Studio on a 240 Hz display no longer reads 80 a second while the scene fills every refresh.

Account

  • The account panel is your profile. It stands you on the scene the graph is playing, with how many scenes you own, have published and the hearts on them; then your credits, your newest scenes, each a press from the graph, and what you have published with where each one stands.
  • Nothing about how you use FluidEQ is sent. Signed in or not, FluidEQ no longer tells its maker how long each scene is on your screen or which audio engine you use, and what was sent before is deleted. What the service keeps is what you do on purpose: the scenes you get, like and publish.
  • What another player played is not kept. A browser tab or another player's song shows on the bar only while that player is playing it, and is gone once it closes, on this computer and on every computer in a Share Audio group; it no longer comes back when FluidEQ starts. Songs whose sound you keep are filed by a fingerprint, never by their title, and the Media page's download log no longer names the files.
  • Sign in with Google. Continue with Google opens Google in your own browser, where you are usually signed in already, and FluidEQ finishes the sign-in when Google sends you back; the panel says it is waiting and keeps a Cancel beside it. Your Google password never passes through FluidEQ. Offered once Google is switched on for FluidEQ's accounts.

What's New and Help

  • Help is drawn from the app itself. Its pictures of the title bar, the Compact player, Share Audio, the look picker, the Forum and the Visualizers pages are FluidEQ's own parts, in your language and your window's colours, so they always show the app you have. The downloadable guide carries the same text.
  • Films of the real app. Fourteen of What's New's slides, and the Help topics they belong to, now have a short film of the feature being used, with a voice saying what it is and how to use it, in your language, and the line being said written under the picture. Watch the video, under a slide's picture, opens its film as large as the window; in Help the film stands after the topic's pictures. Nothing is fetched until you press play, and a film that cannot be loaded leaves the slide and the topic as they were. The films come from fluideq.com.
  • What's New shows the real Output panel for "Every output keeps its own sound", in your window's colours and language, switching from headphones to speakers with each one's profile.
  • What's New has a slide for My curves, drawn with the page's own list: a curve saved from Squiglink lands at the top and plays. Help's headphone topic covers the Squiglink tab and My curves.

Forum

  • Help's button counts what is new in the Forum, in yellow (99+ past 99), and so does the Forum entry in its menu, so new topics and replies are never missed.
  • GitHub Issues in the Forum. A row of their own lists every reported problem, open and closed, with its labels; open one to read the whole thread, and Comment on GitHub answers it there. New activity on an issue counts in Help's yellow number too. Everyone can read them, signed in or not.
  • Clear or switch off the count. Under Forum in the Help menu, Mark all as read clears it, and Turn off Forum notifications takes the number off Help's button for good, until you turn it on again.
  • New for you. The forum's first row lists only what is new to you — topics you have not opened and replies since you last looked — from every board, latest first. Boards with something new carry a dot, new topics and replies are marked, a chip in a thread jumps to the first new reply, and Mark all as read clears them.

Fixes

  • Output names in any language read right: a name in Chinese or any other script came out as question marks and boxes.
  • A Share Audio group of one says "1 computer", in every language.
  • Reset all settings also sends every page back to where a new install opens it and turns the graph back on, and it no longer changes your language. The window keeps its size and place.
  • Dragging a band is heard as it moves again. Each step used to wait until every output's settings had been written, so with many outputs the sound trailed the hand by seconds.
  • Curve smoothing starts Off: a headphone correction plays exactly as it was measured until you choose 1/12 or 1/3 octave in EQ mode.
  • The play triangle on What's New's Watch the video key shows in the key's own colour; it was drawn in the dark of a filled key and could not be seen.
  • A bug report says how hard the FluidEQ Engine works on each output, as an Engine load line, so a sound that cuts out on a busy computer can be told from one that never ran.
  • Switching to an output no longer brings back a preset curve that does not match the preset shown on its picker, which outputs set up before 2.1 could carry.
  • Stopping the music no longer slides the live graph to the right: what is left of the sound falls where each frequency was heard.
  • The stars, glints and out-of-focus lights of the 3D worlds are soft dots again; each was drawn as a glowing ring round a dark centre.
  • Aurora's snowfall, and the rain, snow and crowds of other 3D worlds, no longer vanish on the compact player when the camera turns away from the middle. Scenes the Studio's AI writes lay theirs out so they show on 2.0 too.
  • The Room plays on outputs running at 88.2, 176.4, 352.8 and 384 kHz; it used to say a head file was missing and pass the sound through.
  • The FluidEQ Engine plays on outputs whose driver lists its effects only one at a time, such as Creative's Sound BlasterX G6.
  • Finding the slot an output's driver reads now steps past the slots another program's effect holds, instead of stopping at the first one.
  • On an output where Voice restoration cannot run, the rest of the DSP rack plays and every edit is heard; it used to keep playing the sound from before the edit.
  • Renaming a profile no longer loads it under the old name, and opening the window no longer reports an error for an output with no profiles yet.
  • Play and pause on the bar, and the Library pausing other players, keep working when FluidEQ's media helper stops by itself: it is started again, and so is the volume slider's.
  • The Media page keeps playing a site's next video after a long gap between two, such as an ad being skipped; it used to close in the gap and the next video never started.
  • The desktop visualizer keeps following the music when FluidEQ's window is minimised.
  • The band menu offers Update engine where it says a band needs a newer FluidEQ Engine.
  • The window's colours under a scene stay that scene's own at any Brightness, only lighter or darker; dimming used to swap them for the scene's night palette (Bloom turned from green to purple).
  • A scene republished a moment ago opens as the new version in the Studio and the gallery; it could come back as the version before it for a minute.

FluidEQ 2.0 is a new window around a rebuilt sound.

The window stands on one open floor, in the colours of a new icon. The theme is one slider, Brightness, from near-black to a light ocean blue, and a Plus visualizer can lend the window its colours or play behind the whole app while the panes float on it as glass. Dialogs and menus share one material, and one switch folds everything into the Compact player — the classic amp, or on the Backdrop a player of glass over the visualizer.

The sound was rebuilt preset by preset. Every music style is a whole chain, its bass, air and width chosen for that style, measured and levelled so switching changes the character and never the volume, and each explains itself: what the style asks for and why every point on its curve is where it is. On the FluidEQ Engine your curve plays exactly as drawn all the way to the top, switching presets crossfades instead of crackling, every edit is heard the moment you make it, and Auto normalize lands on the new level in one step. Dimension now widens even a mono record.

The graph reads the sound the way studio tools do: twelve professional views, among them Analyzer, Spectrogram, Third-octave RTA, Peak & average, Waterfall, Stereo & loudness and Oscilloscope, on scales that hold still, with the live sound measured 80 dB deep and down to 10 Hz, so what you see is what is playing.

Every one of the graph's forty looks now draws on your graphics card at your screen's own rate, Plus visualizers can be real 3D worlds that dance to the song and throw sparks from your mouse, the Room has twenty-four rooms, free for everyone; games get their own sound; the Library keeps even a huge collection on disk and opens it at once; the Studio opens with a free trial; and Help answers questions in your own words.

1.7.5 and 1.8.0 were never released; everything they carried is here.

A new look

  • A new look for the whole window. The panes no longer sit in cards: they stand on one floor, parted by fine lines. On the equaliser's pages the graph stands at the top, under the section tabs and the title, and the bands' sliders share its width, each in a place of its own: dragging a band's frequency on the graph leaves its slider where it is, and two bands trade places only when one passes the other. The header's wave lost its box and keeps only the app's rounded corners, the menus lost their rainbow edge, and the tooltips are FluidEQ's own. The boxes on every page — the DSP stage, the empty Library, Share Audio's roles, the karaoke pitch lane and the whole Karaoke Maker, Studio, the gallery — lost their grey fill too and stand on the floor as the panes do, and every dialog is one material: the menus' floor at 95%, the same fine edge and corner, and no rainbow round it.
  • The theme is a slider. Light and Dark became one slider, Brightness, from near-black at the left to a light ocean blue at the right, and every step between them is a theme: the window follows the thumb as it moves. The old Ocean theme stands three quarters of the way along, which is where anyone who had chosen it finds it. The menu behind the pulse icon has it as Brightness, as Window colours does, and the Compact player's menu as its theme, Dark to Light.
  • The colours match the icon. The accent is the cyan at the middle of the icon's wave, at every Brightness alike, and the filled buttons, the switches, the power key and the bar under the open tab run aqua to azure the way the wave does.
  • A new app icon. The wave, glossier, on a dark tile, in Lagoon's colours with an edge to match.
  • The title is alive. When FluidEQ opens, the wave in its logo draws itself and the name rises letter by letter; after that a pulse of light runs along the wave every few seconds. "Fluid" is white, and "EQ" and the logo's edge and wave are in the window's colours — the accent, a visualizer's colour or the Rainbow palette; the dialogs keep the app icon as it is.
  • Rainbow mode is on from the start, and it is Lagoon. Aqua, cyan, sky, azure and a soft periwinkle — the icon's own colours — where it used to run red to violet, and there is nothing to unlock any more: the switch on the signal, the What's new slide and the offer in the Support dialog are gone, and Normal or Rainbow is a switch under the sliders in Window colours and under Brightness in the app menu, and a pill on the wave in the title bar. The window takes a little of it too: the panes lean toward its sky.
  • Window colours, by name. The button beside Auto on the graph says what a Plus visualizer does to the window — Theme, Colours, Ambient or Backdrop — and opens all four, each with a line on what it does, under three sliders. Brightness is the theme's own slider: with the theme it goes from near-black to the light ocean blue, and while a visualizer lends the window its colours it goes from their darkest, never black, to their lightest. Transparency is how much of the visualizer shows through the panes on the Backdrop, and stands dimmed in any other mode. Both are marked at a quarter, a half and three quarters, and the thumb falls into each mark. Daylight is the visualizer's time of day, from night to full day, and two switches under the sliders can set it for you: Daylight follows Brightness, on in a new install, and Daylight follows the clock — night until 05:30, full day from 08:00 to 17:30, night again from 20:00. With both off, Brightness changes only the window and the scene keeps the Daylight you gave it. A new install opens at 50% Brightness and 50% Transparency, with a Plus visualizer on the Backdrop; the full app and the Compact player each keep their own, and Brightness follows the hand without the window holding it back.
  • Pointer sparks. Move the mouse across a Plus visualizer and it throws what it is made of — sparks, petals, snow, embers — in a trail behind the pointer, with a burst wherever you click. They are on in a new install, and one switch turns them off or on for every visualizer: Pointer sparks, in Window colours under Rainbow mode. The Studio's stage always shows them, so a scene's maker can shape what it throws.
  • Backdrop. The fourth mode puts the Plus visualizer behind the whole window, the graph still its frame, and the panes and everything on them let it show through. Menus do not: every menu stands at 95% whatever the Transparency, where one opened from a see-through pane used to go see-through with it.
  • The Backdrop floats the panes. On the equaliser's page the title bar and the player bar run to the window's edges, and the side column, the sound panel and the bands stand on the visualizer as rounded panes of glass with room around them. Folding the sound panel leaves only its button, and the graph and the bands take the room it gave up.
  • The Backdrop stays behind every page. With a Plus visualizer on the Backdrop, going to the Library, Karaoke or any other page keeps it playing behind the window; in Colours and Ambient it rests off the graph's page, so nothing is spent drawing what nobody can see. Switching between Colours, Ambient and Backdrop, or into full screen and back, moves the same visualizer instead of loading it again.
  • A Plus visualizer lends the whole window its colours. Five are read from the visualizer itself as it plays. In Normal the buttons, the switches and the "EQ" of the title take its first colour and what is switched on its second; in Rainbow all five run across the bands, the title and the lit controls. The app's icon stays its own.
  • A Plus visualizer fills the graph. It runs up behind the section tabs and the title, and down to the divider, with no band of floor around it.
  • Cards in the window's colour. The Library's tiles, its song list and Up next, each game's row, the outputs in the engine's config, the headphone picker, Share Audio's roles and the Studio's AI card are filled now — a gentle step above the page, never lighter than a pane, with the accent glowing in from the corner, in the theme's colours or a visualizer's — where they were outlines or slabs of grey slate. The small graphs (the headphone correction, the driver sketch, the DSP's plots, the Share Audio monitor) stand in a darker well, so their lines read on the darkest thing around them. What holds them, like the grid round the tiles, keeps its outline, and a control that cannot be used is an outline with dimmed text instead of a lighter slab.
  • The expanded graph is square. Its card and the picture under it have square corners, as the column they cover does.
  • The graph's paper is plain. The shaded bands across its top and foot, past ±20 dB, are gone; the ±20 dB rule still marks where the scale tightens.
  • One knob, and sliders that feel right. Every knob in the app is the same lit ring with its reading under it, and the reading can be typed into. The EQ sliders have round or rectangular handles, as you choose, a core lit with each band's level, a detent at the centre and a snap at 0 dB. A new install gives the full app rectangular handles and the Compact player round ones; a player that never had a choice of its own takes the full app's.
  • Every dialog on one frame. Every dialog and notice stands on the same frame with the same close button, and the keyboard stays inside a dialog while it is open.
  • The sound panel folds to a rail from a button at its top left, the settings pages and the panel's sections stand on soft cards, as does the graph under Plus and the forum, lined up with theirs, and every shade's lines, text and outlines were set for contrast — the darkest end is truly black. The Bands page has its title back, and the curves a band layout carries are in one Curves dropdown on its bar.
  • Small things made clearer. The info mark beside a preset's name and on the DSP page glows like the switches, Game mode's pad reads as a gamepad, and a visualizer's icon no longer sits on a black square.
  • On a Mac, FluidEQ looks like a Mac app. The Mac's own close, minimise and zoom buttons stand in the title bar and in the player's strip, with the Mac's corner and edge, text drawn the way the Mac draws it, and the green button and a double-click on the title bar doing what your Mac is set to do.

The Compact player

  • The Compact player. The switch beside Help in the title bar turns the window into a player a few hundred pixels wide: the song and its clock at the top, then decks you open and close — the equaliser, a visualizer and Up Next — the window growing and shrinking by what each takes. The equaliser comes along whole: presets, the band layouts, EQ mode, Smart EQ, Clear EQ, and the bands or the Bass, Mid and Treble dials. The visualizer plays any of the graph's looks and fills the screen on a double-click. Up Next shows where you are in the queue and the time left in it, and music files dropped on it join the Library and the queue in one gesture. The player remembers its own size and place, can stay on top of other windows, and folds to one line on a double-click of its strip; its menu opens the full app on any page. If it ever ends up off the screen, FluidEQ's icon in the tray has Recover the window.
  • On the Backdrop the player turns to glass. With a Plus visualizer on the Backdrop, the Compact player is the visualizer itself, edge to edge, with the song in large type, the controls in a glass dock, and the equaliser and Up next in a sheet at its foot; in every other mode it is the classic amp, with its LED clock and lamps. The full app, the classic amp and the glass player each keep a window size of their own, and a switch from one player to the other keeps the window hidden until the other has drawn. The classic amp's colour key opens the same Window colours menu as the graph's button.
  • Where the sound goes. Under the song one strip shows where the sound comes from, FluidEQ, and the output it reaches, with the output's rate and layout on the FluidEQ Engine, and a light runs along it while sound is playing. The source opens its page, FluidEQ opens the equaliser, and Game mode stands beside it with the delay it cuts.
  • The Compact player keeps a theme of its own. Its menu has the theme's slider too, remembered apart from the full app's, so the app can stay dark while the player sits light beside your work.
  • The player's visualizer has the graph's switches. Changing looks by itself, a scene's colours on the window, the desk lights and the desktop background, in the same order as on the graph: beside the picture when the player is two columns wide, and in the corner of the equaliser's screen when it is one.
  • The volume is your computer's own. FluidEQ's separate volume level is gone: the slider on the player, on both bars and in Karaoke is Windows' volume, so it sets the level of everything the computer plays and stays where you leave it. Every player in FluidEQ plays at full level into it.

Presets and the equaliser

  • Presets that sound like the music. Every preset was measured again and levelled, so switching changes character and not volume. Each music style is a whole chain — its bass, its air and its width chosen for that style, and only where they suit it. Music, Movie and Gaming each have a Room copy, with five measured bands in front of the Room so the chain sounds like itself through it (a record leaves a room about 3 dB louder and 3 to 4 dB darker than it went in; those bands are that difference).
  • Every music style explains itself. Each genre preset has notes: what the style asks for, why each point on its curve is where it is, and what every effect in its chain does or why it is left off, with how loud it plays against DSP Off. They open from About on the DSP page, from the i on the preset's chip under Also applied, and beside the list while you point at a preset in either picker, and the points they explain stand on the preset's curve on the graph. Every preset was checked against its notes and does what they say: a style mixed with its bass in the centre plays it centred below the point that style names, and no curve lifts where its style keeps things flat.
  • A Ballad preset, with the pop styles: the slow song of every catalogue, pop, rock, soul and Latin alike. The voice stays close and clear — a little warmth and presence, the low-mid mud and the sibilance eased rather than brightened — the bass is centred, and a light, slow ceiling keeps the quiet verses quiet. Nothing is added that the record does not have: no harmonics on the esses, no invented sub. Its notes explain each choice, as every style's do.
  • None is one choice, first in both pickers. Picked on the equaliser or on the DSP page, it takes the whole preset away — its tone and every stage of its chain — and it stands above everything in both lists, the chains you star included. Reset on the DSP page puts the Default chain back, tone and all, instead of an empty rack that sounded like None.
  • A preset's tone plays in the main equaliser. It used to sit inside the rack, ahead of its compressor and limiter, where on the FluidEQ Engine Pop against Metal was barely a difference at all. It is now the Preset layer of the main equaliser on both engines: it shows on the graph as a curve of its own, named after the preset, with a strength you can turn down, and its × takes the whole preset away. The rack keeps what supports the tone.
  • Width you can hear, even on a mono record. Dimension's Spread now makes width out of the middle of the mix, above the bass, so a mono recording or a voice dead centre opens out too; before, the stage could only stretch the width a record already had, and a mono one stayed mono. Someone listening in mono still hears exactly the same, and the bass stays in the centre. The Movie preset and the Dimension card's Gaming profile keep the centre focused, so dialogue and a footstep straight ahead stay where they are.
  • Shape a bell across several bands. Select bands on the graph, hold Ctrl and drag one of them: it becomes the top of a bell and the others follow its curve, less the further they are from it, without moving along the frequencies. The bell is as wide as that band's Q, so Ctrl+scroll widens or narrows it, and letting go of Ctrl mid-drag moves the group as one again. A Ctrl-click on a selected band still takes it out of the selection.
  • Smart EQ measures the song, not what FluidEQ has already done to it. It used to listen after every layer and the DSP rack and keep correcting its own previous answer, so the same song never got the same curve and the rack's colour was "fixed" again on top. It now hears the sound before FluidEQ touches it: the Library's own input while the Library plays, and the playing app's sound otherwise. The same song always gets the same correction, a new song starts fresh, and the correction is replaced rather than piled up. The graph's gap bars and countdown are gone, and the Smart EQ button is never greyed out; its bubble says when it cannot hear the source.
  • Bass, Mid and Treble across the whole rack, the way an amplifier has them, on the main equaliser whenever no band is selected. Each moves its own third of the spectrum; Ctrl+click puts that third back to flat. They are a curve of their own — drawn in their own colour, with a chip under Also applied, and saved with the profile — so turning them never rewrites bands you tuned by hand. The equaliser now opens with nothing selected, showing the whole rack.
  • A low cut and a high cut on the Tone panel, either side of Bass, Mid and Treble: off, 12 or 24 dB per octave, at 20 Hz and at 20 kHz. Clear EQ turns them off with the bands and the Tone.
  • The equaliser plays the curve you drew, all the way to the top. On the FluidEQ Engine a band near the top of the range used to play weaker than the graph showed — nearly 4 dB short at 20 kHz on a 48 kHz output. It now plays as drawn. The EQ mode menu's Treble row keeps the old way as Classic, for your EQ and for headphone corrections separately — Classic is how Equalizer APO builds a band and how AutoEQ tunes a correction — and the graph draws whichever one is playing.
  • Auto normalize lands on an edit's level in one step. Applying a curve or moving a band used to drop the volume by the most it could ever need and climb back slowly, still moving about 2 dB ten seconds later. On the FluidEQ Engine it now works out the level the new sound needs from the last ten seconds of music, moves there in one step, and only fine-tunes after that.
  • Edits are heard the moment you make them. On the FluidEQ Engine a band you drag, a curve you apply or a preset you pick used to wait while that level was worked out, so a drag was heard three or four times a second and a preset arrived half a second late. The change now plays straight away and its level follows a moment later, gliding into place — which also took out the small click a preset switch made when it turned the volume down. On a Room preset an edit no longer rebuilds the room either, so a drag there is as quick as anywhere else.
  • The Maximizer lets go without a tick. At the end of every release it used to jump the last fifth of a decibel in one step: a faint tick after each peak it held down, clearest a moment after switching to Punch. It now eases the whole way.
  • Linear phase on your EQ and on the curves shares one delay, instead of adding the two together.
  • A twenty-band quick layout, and every layout moved onto the international standard's frequencies: a band labelled 64 Hz is now the 63 Hz everything else means by it. Bands also open as wide as their spacing — a thirty-one-band rack no longer has three bands playing the same note, and a six-band no longer leaves holes between them.
  • The EQ sliders keep up with the hand. A drag moves the slider at once and the graph follows without holding it back; the engine gets the newest value as soon as it is ready for it. Only the slider being dragged redraws, where every band and control on the page used to.
  • The equaliser shows the output's rate beside its name, as the DSP page does: the rate Windows runs the output you are listening to at, read again when the output changes or you come back to the window.
  • The FluidEQ Engine has a new build. The copy on your machine is offered an update the first time you run 2.0.0: one press in the notice, and Windows restarts its audio itself. Until you take it the equaliser keeps working and the DSP effects stay off, which is what the "Update engine" card says.

Games

  • Every game its own sound. A chain per game — from Steam, Epic Games, EA, GOG, Ubisoft, Battle.net and Xbox, or any program that is open — switched the moment the game comes to the front, and kept until the game is closed, however often you alt-tab. A card on the desktop — on the screen FluidEQ is on, in the window's own colours — says what was loaded and another what came back. While a game holds the sound, the player bar names it. The equaliser's fourth pill is named for what it holds — Game presets.
  • Game mode, and the delay you can see. Game mode cuts the delay FluidEQ adds, for aim and for voice chat, and the Gaming presets turn it on. The EQ and DSP pages now show the delay the chain adds as measured, broken down by the effects that add it, and Linear phase says what it costs.

The Room

  • Thirteen new rooms, told apart by measurement. Twenty-four in all, each with its speakers where that place would put them and its own air, and none of them another one under a different name: every pair is measured against every other for level, for how wide it sits at the ears, for top against bottom, and on a surround stream for the centre, a side and a rear against the front. Three rooms written that week measured as copies of ones already there and were taken out again.
  • The Room's page hides nothing. The rooms are in the same picker every other stage uses, under Featured, Classic rooms and Yours; what a room is made of is on the page rather than behind tabs; the room stands on the left with six bands beside it, and the speaker you pressed has its own pane from the moment you press it, with its angle as a number you can type.
  • The Room is not part of Plus. Shaping a room, saving one, fitting the head by ear: all of it is open to everybody.

The graph and the visualizers

  • The graph opens where it is used. Only the equaliser's Bands page opens with the graph, above the bands. Every other page — the equaliser's Presets and Convolution, DSP, Online Media, the Library — opens without it, and a graph switched on there stands under the page.
  • The graph shows the whole spectrum, on scales that hold still. With the grid on it starts an octave below 20 Hz, where a low cut does its work; with the grid off it is trimmed to 20 Hz – 16 kHz, where records have sound, so a full-screen picture has no empty strip down the side. The equaliser's ±20 dB scale no longer stretches when a curve runs past it: the last tenth at the top and bottom holds everything out to ±60 dB, compressed, so a curve that goes further — the final output under a deep preamp or Auto normalize — is drawn whole instead of cut off flat. The live sound has a scale of its own on the right, 80 dB deep as a studio analyser draws it, so the top octave and a high cut show. Frequencies are labelled 10, 20, 50, 100… the way analysers print them.
  • Double-click the graph to fill the screen, Ctrl+double-click to fill the window, and double-click again to come back from either. Help lists the graph's keyboard shortcuts.
  • Twelve ways to read the sound. The graph's plain drawings gave way to the views studio tools use: Analyzer, Before & after, Spectrogram, Third-octave RTA, Peak & average, Waterfall, Stereo & loudness, Oscilloscope, Mid & side, Note spectrum, Energy bands and Phase history. The Waterfall is a wireframe surface of the last second and a half, gliding back from a front edge that follows the music. A look you saved on one of the old drawings still draws.
  • Every look on your graphics card. All forty of the graph's looks now draw on the engine the Plus scenes run on — the same drawing in the pane, the expanded view and full screen — at your screen's own refresh rate, where the standard looks used to be held to 30 frames a second. Changing looks fades from one to the next, and a look keeps its peaks and particles through the hand-over.
  • Eighteen new visualizers under Scenes. LED wall, Glass towers, Tide, Halo, Synthwave and Horizon; LED bars with square segments, Neon bars, 3D bars and Spectrum wave; and Pixel bars, Spark bars, Glitch bars, Halftone, Bouncing dots, Falling blocks, Fiber optics and Afterglow. Each moves with the bass, the beat and the treble separately, keeps its shape upside down and mirrored, and follows the look editor like every other look: colour by frequency, level or loudness, pieces and gap, filled or outline, opacity, line width, texture, lit peaks and glow.
  • The looks and the level meter wear the window's colours. A look with no colours of its own draws in the accent and its lighter and darker shades in Normal, and in the Rainbow palette — Lagoon's or the Plus visualizer's — in Rainbow. The level meter and the look editor's colours follow the same way. The meter's ten styles are drawn flat and clean, and the title bar's wave draws crisp bars on the screen's own pixels.
  • Plus scenes turn from night to day with Brightness. Every FluidEQ scene has a day look, and slides toward it as the window gets lighter.
  • A visualizer keeps its subject in view in a narrow window. In the Compact player and any other narrow window, a Plus visualizer slides its picture to keep what it is about on screen instead of squeezing it.
  • 3D worlds run lighter on laptop graphics. A world no longer spends a quarter of each frame on work Intel graphics did not need.
  • The wave's height and position are kept per view. Set in the graph's View menu, they are remembered separately for the pane, the expanded view and full screen, and for each Plus visualizer. The Compact player and the desktop backgrounds draw a visualizer with its full-screen setting.
  • Another player's cover behind the graph. When the sound comes from Spotify, a browser or any other player, its album art fills the graph behind the wave in the expanded view and in full screen.
  • Visualizers in real 3D. A scene can now carry a 3D world — shapes and models, materials, lights, reflections on wet floors and water, glow — standing in front of its painted sky and moving with the music: the kick, the snare and the hats each their own way, gathering in a build and breaking on the drop. It goes through the same smoothing, FSR upscaling and brightness limiter as every scene, everywhere a scene plays — the graph, the desktop, the lights, the Studio and the gallery — turns with your mouse where its maker allows, and hands its graphics memory back while the window is covered. In the Studio a world is a file beside the scene with its own GLSL and models, and opening one of FluidEQ's scenes writes its world out for you to look inside. A FluidEQ from before this plays the same scene without its world.
  • Scenes can dance to the song. A scene now hears the tempo and where each beat and bar falls — a clock that lands on the beat instead of answering after it — the kick, the snare and the hi-hats apart, how intense the song is, when it builds and when it drops, where the music sits between the speakers, and the singer's voice. The AI prompt teaches all of it, so a scene written from one line can step on the kick and leap at the drop.
  • Turn a 3D scene with your mouse. Where its maker allows, drag a scene to see it from another side, Ctrl+scroll or pinch to move in and out, and press Reset view for the view it was made with — on the graph, in the Studio and in the gallery. On the graph a plain drag still selects bands; the right or middle button turns the scene there. A scene can also answer where you point and tap.

The desktop, Online Media and Windows

  • YouTube, YouTube Music, Twitch and Suno in FluidEQ's colours. Match FluidEQ's colours, in Online Media's toolbar, puts a site's dark page in the colour of the card it stands on, its cards, menus and top bar a step lighter as the app's own are, and its search field and playlist in the app's outlines. It is on your screen only: text keeps its own colour, and videos, pictures, ads and logos are never changed. It is on in a new install and can be switched off, and works in each site's dark mode.
  • A Plus visualizer behind Online Media. With a Plus visualizer on the graph and the site's colours matched, the video's own full screen plays the visualizer behind the page: the chat, the comments and the suggestions stand over it, and the video stays solid on top.
  • A desktop background keeps its place. It plays on while its replacement is prepared, pauses while windows cover its whole monitor or a game is in front, gives the desktop its own wallpaper back while it waits on battery, and follows the visualizer's settings from the window as you change them.
  • The desk lights stay on while FluidEQ runs. Minimised, covered or behind another app, your devices keep following the scene. They stop only when you switch them off, pick a look that isn't a Plus one, or close FluidEQ.
  • The desk lights come on as FluidEQ opens, music or not. They used to wait for sound, and stayed dark while Windows was still opening or restarting its audio.
  • A monitor can follow the graph. Switch on Follow graph when you put a visualizer on the desktop from the graph or the player, or for each monitor in the Manage dialog, and that monitor changes to whichever Plus visualizer the graph shows, whether you picked it or the graph changed looks by itself. When the graph shows a standard look, the monitor keeps the last Plus visualizer it had. Follow graph starts on when the visualizer you put on the desktop is the one the graph shows, and off for any other, so the one you picked is the one that plays. A monitor that follows the graph wears a small link mark in both dialogs, and turns from night to day with the window's Daylight: Brightness, the clock, or your own.
  • The desktop dialog chooses the monitors for you. A visualizer no monitor shows yet starts with your main monitor chosen, and one already on the desktop starts with its own monitors chosen. The first monitor chosen plays the visualizer live in the dialog's picture, with the wave it will be set with, before you set it.
  • The window gives way on a small or zoomed-in screen. The graph's buttons fold onto a second line instead of running off its edge, and the title bar lets its level meter go first, then moves Help and the Compact player switch into the actions menu; the tabs and the window's own buttons always stay.
  • Start with Windows. A switch in the actions menu, under Animations, starts FluidEQ when you sign in. It needs no administrator, and it says so when Windows' own Startup apps has FluidEQ switched off.

The Library

  • The Library keeps your music on disk, not in memory. Songs, albums, artists and folders are read a page at a time from a database on disk, so a library of tens of thousands of songs opens at once and costs little memory. The library you already have moves in by itself the first time 2.0 starts.
  • Search inside a folder finds the whole library. Matches in the folder you are in, and the folders under it, come first under "In ". Everything else that matches follows under "Everywhere else".
  • One Back, and a trail of where you are. Every shelf and view of the Library shows Back and the path from All music down to the folder you are in. Switching views inside a folder never leaves you without a way back to the top.
  • The player bar keeps the last thing you played. After a restart, a closed browser tab, the other computer leaving Share Audio, or Stop, the bar and the Compact player go on showing the last song — from the Library, another program or another computer — until something new plays. They used to say Nothing playing. A click on its title goes to the page that played it.

Plus and the Studio

  • The Studio is part of Plus, and opens with a free trial. Fifteen days, no card, nothing charged when it ends — for an account created on or after 20 September 2026, the day the offer opened. Somebody without Plus meets a page that says what the Studio is and how to get in, and names the folder any project of theirs is sitting in, whole and untouched, for the day they come back to it.
  • Publish a scene, keep FluidEQ free. A scene of yours that is approved earns the next month of Plus. The account panel shows the month you earned, counts down its last week and says so on the day it ends: nothing renews quietly and nothing is charged. A maker whose month has run out can still publish the next scene, and keeps one project on the bench.
  • Every member's scene is held for approval before anyone sees it, and a new version is published above the highest number that scene was ever out at, so a number can never come back around. A scene you find wrong can be reported from the looks menu.
  • Your AI can see your scene, and hear the song. Copy the AI prompt in the Studio and your AI assistant — Claude Code, Codex or another that speaks MCP — connects to FluidEQ on this computer by itself. It can draw your scene at any shape and moment, with music or in silence, read why a scene cannot play, and hear how the song you are playing moves — its sections, builds, drops, drums and tempo — as numbers, never the sound and never which song. The switch is on the Studio's card, and nothing of it leaves your computer.
  • The Studio's meters follow the rhythm and the song: the tempo and how sure it is, a lamp each for the kick, the snare and the hats, a spin the music winds up, intensity, build and drop, the voice and its note, and where the music stands left to right. The test music plays a song's shape — a verse, a build and a drop, the kick on one and three and the snare on two and four — so a scene's answer to each can be tried.
  • The Studio keeps its stage in view. The work on a scene sits in tabs under the stage, the stage offers Full screen and the grid, and the Studio's window colours are Theme, Colours and Ambient.
  • The Studio shows a scene behind the Compact player from a button on the stage, so a maker sees how it frames in a narrow window. Each "things in the window" slider shows the icons of what it moves, and Brightness and the wave's height snap to 25, 50 and 75%.
  • Add to my looks says what it did: a notice under the button says whether the look was added or updated, and the Studio's notices show above its work again.
  • The Plus terms, third edition. They now cover the approval every published scene waits for, the free trial, the month an approved scene earns and the one Studio project a maker keeps without Plus, that a scene can carry 3D models, and what an AI tool on your computer can see once you let it look at the Studio's stage. Members who agreed to an earlier edition are told once, in a small card, what changed.

Help

  • Two videos. The guide's first chapter plays a four-minute video of FluidEQ's first steps, and the FluidEQ Plus chapter a tour of Plus: Visualizers, the Studio from an idea to a published scene, the desktop visualizer, Dynamic lighting and the Leaderboard. Both are narrated in English with captions, streamed from fluideq.com when you press play and never before; the site shows them too.
  • Ask the guide in your own words. Help's search ranks what it finds, forgives typos and plurals, knows the words people use for things in all ten languages — "no sound", "limiter", "wallpaper" — and searches the English guide alongside your own, so an English word finds its chapter too. Each result quotes the passage it found with the words marked; the guide takes you there and rings the control on its picture, and Enter walks to the next match.
  • Help shows the 2.0 window, and has a chapter for the Compact player. Its pictures were taken again at the middle of the Brightness slider, so one set reads on a dark window and a light one. Help itself is now the book in the title bar, beside the actions menu.
  • Reset all settings. A new item in the actions menu puts every setting back the way a new install starts. Your EQ, presets, curves, DSP, outputs and audio engine stay as they are.

Fixed

  • A band's fader moves its own band, and stays where you leave it. Turned with the mouse wheel, the arrow keys or a screen reader, without a click on the band first, a fader moved whichever band was selected, or none at all. And the last step of a quick movement could be lost, leaving the sound and your saved profile a step behind what the fader showed. Each fader now moves its own band, and the band plays exactly where you left it.
  • The graph fades in, and nothing in it slides. Opening a page with the graph, the grid and the curves used to glide in from the edge, and leaving full screen squashed the visualizer through each size the window passed on its way back. The graph now stays out of sight until the window has its size and fades in there, whole; the grid and the curves take any new size at once. Showing and hiding the graph still slides the band sliders.
  • Your Plus visualizer comes back after a slow moment. When the computer could not keep up with a visualizer for a while, or it failed to start, the graph showed its plain look for the rest of that session — and kept the plain look for good. The next launch tries your visualizer again.
  • Every hint is FluidEQ's own, even one that appears under the pointer. A button that gains a hint while the pointer rests on it — the Gallery's Add, which becomes Remove once pressed — showed Windows' own tooltip; it is the app's now, and after a press it waits until you move on.
  • A visualizer on the desktop changes with a crossfade. Switching it, following the graph's automatic switching or saving your scene in the Studio used to blink the screen dark and cut to the new one; the old picture now stays until the new one has drawn and fades into it.
  • The Studio's edges are solid. The Plus side rail, which opens over the Studio when folded, is the menus' material now instead of see-through; the buttons pinned at the foot of the Studio's controls hide what scrolls under them; and the bar across the top of the Studio is the window's own colour instead of a lighter band.
  • The EQ opens with its sliders at full length on any screen. On a large screen the bands under the graph used to open squeezed to their shortest, because the Tone dials there are bigger; the first height of that pane is now what the bands need, until you move the divider yourself.
  • The wave in the title bar follows quiet music. A song played softly, or at a low Windows volume, used to lie almost flat on the middle line and jump from frame to frame; it now fills the strip and glides at any volume. The flat line means real silence, and only once it has lasted half a second, so the gap between two notes no longer flashes it.
  • A Plus visualizer's colours arrive as you choose it, not after it has loaded — even the first time, before it has ever been measured.
  • Automatic switching gives every visualizer its whole turn, counted from its first frame rather than from the moment it was chosen, so a slow one is never passed over before it has shown.
  • The equaliser's sliders are never cut off at the sides, with the grid on or off.
  • A rack with every effect off plays the sound as it came. Two things inside it were never shown on any card: a three-band compressor that every factory preset switched on, and a final limiter with a low-cut filter that was always on. Both are gone. Switching DSP on with nothing on no longer changes the sound, and a loud record no longer lights clipping warnings it did not earn; the Master card shows the true peak instead. Where a chain has no Maximizer or Master ceiling it can now peak above full scale like any equaliser, and Auto normalize keeps the headroom, as it does for the main EQ.
  • The Maximizer holds its level instead of pumping. On a dense record its gain dipped on every kick and sprang back before the next, so the loudness swung with the beat. It now settles at the reduction the record keeps asking for and stays there, still catching every peak the moment it comes: the swing is about a third smaller, and the ceiling is exactly where it was.
  • The queue no longer skips a song at the end of a crossfade, Next is instant, and the seek bar no longer jumps back and forth after you drag it. Getting the next song ready used to reset the effects under the song that was playing twice a track — a small hiccup mid-song — and no longer does.
  • Up Next keeps the order you give it. A song dragged to a new place went back to its old one when the next song started, and a song taken out came back; and after taking any song out, the whole queue was gone at the next launch.
  • Switching to the player and back no longer shows a stretched window, and neither does the player's picture going to full screen: the window changes size out of sight and comes back drawn, full screen fades in from black, and a view that measures the sound keeps its grid there.
  • The player's visualizer, two columns wide, reads like the graph. A view that measures the sound shows the graph's grid and scales there, and Ctrl+G shows and hides them as it does on the graph; the level meter runs across the whole picture; a view's legend sits under the picture instead of behind the look picker; and full screen has no frame around the picture.
  • The output level meter no longer slows the window. It printed its scale afresh on every frame, about a quarter of a processor core's work; it now prints it only when something on it changes.
  • The graph's curve chips are in the order of the chips under Also applied: the headphone correction first, then the convolution, the preset, the Tone and your EQ, with the final output after them.
  • The Compact player recovers from an error. Its recovery screen said it was reloading and never did; it now reloads, and comes back as the player.
  • The Studio's colours stay on the Studio's page. Going to the player or another page from the Studio, the window takes the colours of the visualizer that is playing instead of the Studio's scene.
  • Scene icons show in the player's look picker, where they were blank.
  • The desk lights switch on the graph shows its light only while it is on. Switched off, it drew the same rays as on.
  • A visualizer looks the same wherever it plays. Your own scene on the desktop was treated as another member's and left a ghost of each frame behind it, while the Studio showed it clean. The graph, the gallery, the Studio, the player and the desktop now play every scene by the same rules. The desktop also follows the music on every frame it draws, where it used to move in steps thirty times a second.
  • Another member's scene that flashes is calmed, not smeared. The flash protection used to blend the previous picture into fast motion, which showed as a double image. Now it only dims the patches that flash, and everything else moves as drawn.
  • FluidEQ fits a screen scaled to 200%, such as a Mac showing it at 1440 × 852: the title bar and the playback bar are shorter, the DSP pages' pictures fit beside their controls, the cover flow shows more songs, and Karaoke's pitch lane keeps its notes readable.
  • Repeat One repeats the song again in the Library.
  • An EQ switched off from its chip comes back on when you clear its bands. Switched off from Also applied and then cleared with Clear EQ, the chip's × or the Squiglink panel, it stayed off for good with the band editor greyed out.
  • Resizing the window no longer throws it into the recovery screen, and its edges no longer show a tint of the wallpaper behind it.
  • The title bar makes room for itself. The tagline and the little creature leave as soon as the signal would have to narrow, in every language. The signal and the meters no longer stay frozen after you come back from the Compact player.
  • Smart EQ's speech bubble no longer covers Game mode, the delay or the page title. It finds room above or below its button at any window width.
  • Each DSP card is one colour. Every box of controls or readings on a card, the phase meters on the EQ and Dimension cards among them, wears the card's own colour with a thin edge, in both themes, instead of two or three shades of slate; the graphs keep their darker ground. The scopes' trails fade into the card instead of leaving the dial's ghost behind them, and the line under the DSP title is one strip with its button at the end.
  • The Denoise graph shows its numbers. It printed {{count}} and {{hz}} instead of how many clicks it repaired and the hum's frequency, in every language.
  • The DSP page's header keeps to one row on a window 1100 pixels wide and up: the presets no longer leave their file actions on a line of their own, and the Dimension card's legend no longer stands across the header.
  • Renaming a profile says why a name was refused in your language, and the Config page calls the preset's layer a preset, not a voicing.
  • FluidEQ says when it is not reaching the output, instead of looking switched on while nothing goes through it, and the EQ page stops taking changes the engine cannot hear, the way it does when FluidEQ is off.
  • A headset connected after FluidEQ started is no longer reported as out of its reach while FluidEQ is processing it. A Bluetooth headset switched to mid-session was marked in red, and the EQ page stopped taking changes, over an equaliser that was working.
  • Switching output keeps the preset you picked. The new output's profile used to bring back whichever preset it had been saved with, so the preset picker named one preset while its chip under Also applied named another.
  • Changes no longer click. Every change the FluidEQ Engine hears — a band moved, the preamp — now fades in over a fiftieth of a second with each band carrying on where it was, where a big move used to click. Dragging a band through 0 dB no longer clicks either, and switching presets no longer crackles: where a preset holds the sound back by a different amount, the sound crosses from the old preset to the new one, and the effects a preset switches on or off glide in and out.
  • A headphone correction plays as published, even where it goes past the sliders' ±20 dB, and the preamp you set by hand now reaches −60 dB to hold it, as Auto normalize already could. A deep preamp is no longer put back to −20 dB after a restart.
  • The engine does less work. With no headphone curve on it no longer spends two thirds of its effort on a filter doing nothing, the headphone curve in minimum phase adds about 13 ms of delay instead of 55, and silence from a paused player no longer costs thirteen times what the equaliser does.
  • The Target button on the EQ page no longer flashes the wrong look for a moment at launch.
  • Leaving a Plus visualizer no longer flashes the screen white. Pressing Back, going to another page or starting a scene on the graph showed one white frame where the visualizer had been.
  • The graph's View menu is never cut off at the top. It opens clear of the bar above the graph and scrolls inside itself when it is taller than the room.
  • Help's pictures point at things, the way a printed manual does. Each control on a picture has a numbered circle beside it and a line to it, with the same numbers in the list underneath, in the app and in the printed guide alike, and the pictures were retaken against the window as it is now.
  • Every language uses the names on its own screen. Labels that stayed in English are translated; Smart EQ, the Studio, the Exciter, Karaoke and the Equalizer APO engine are called in every sentence what the screen calls them; Portuguese is Brazilian Portuguese throughout; Russian writes decibels as дБ; and the theme picker calls its two themes Light and Dark.
  • "Voicing" is gone for good. The equaliser's picker became Presets, and a preset's curve on the graph, its chip under Also applied and the driver list's hint say Preset too.
  • The engine finds the place a sound card actually loads it from. Windows keeps an effect in one of eight places and which one a driver reads is written down nowhere, so FluidEQ steps through them on the machine itself. It was stepping over three of them, and on a Bluetooth headset it came to rest in a place that endpoint is never read from.
  • The Room is as loud at 192 kHz as at 48, a flush of Windows' audio pipeline no longer puts a hole in it, and the room you draw is drawn to scale: a speaker further away than the room is wide stands outside its walls, where it is. Saved rooms keep every speaker's setting, mutes included.
  • The Master's loudness target works everywhere, not only on tracks played from the Library, and its readout says why it reads +0.0.
  • The DSP page comes back to the stage you left it on, instead of always the Normalizer, and no longer shows a "System-wide" pill that told you nothing the page did not.
  • Eight controls that took an answer and did nothing with it are either wired up or gone, and three more are no longer drawn on the screens that cannot read them.
  • Shared audio arrives exactly as it was recorded. The computer you share from sends its sound untouched, and the computer with your headphones or speakers applies its own EQ and effects to it once. Share Audio no longer asks the sender to pick Music or Game/Video.
  • The leaderboard lists every place from first down. The three on the podium are in the list under it too, their ranks in gold, silver and bronze, where the list used to start at fourth.
  • Plus given as a gift is not described as a subscription, and a maker with another earned month waiting keeps Plus the moment one runs out.
  • Sending more scenes than a month allows says so, and when you can send again; a server answer nobody can read no longer shuts the Studio. A member with as many scenes out as one member can have is told that, and to unpublish one, instead of being asked to fix a scene with nothing wrong in it.
  • The Studio's meters keep time with the music. They move the moment the sound reaches you, where they came a drawn frame behind it; Bass follows the bass line and the kick, where a man's singing lit it; the beat lamp flashes on every beat once the tempo is found; and the stage keeps one steady frame rate instead of stumbling between two.
  • A scene's colours reach the rest of the window, and the Library stops blurring FluidEQ's own scenes as though they were a stranger's; on a scene's page the picture fills the height of the panel beside it instead of leaving an empty band under it.
  • Signing out asks through a clear card, the Animations switch takes effect at once, and the player bar no longer leaves an extra copy of its helper — the one that follows what other programs play — running in the background after the window reloads, where it could also say Nothing playing over a song another program was playing.
  • Double-click a playing song to start it again, in the Library's list, grid and Cover Flow and in the Compact player's Up Next. A double-click in the Library's Up Next no longer skips to the song after it.
  • Up Next stands still while the music is paused. The bars beside the playing song kept moving while the Library was paused, even when another program was the one playing.
  • With the grid off, the drawing stands on the graph's floor, instead of keeping the grid's margin as an empty strip above the band sliders.
  • Braid is back. It had been folded into the Analyzer with the plain drawings.
  • Full screen shows the graph on a zoomed-in window, where below 940 pixels wide it showed an empty window; and the Analyzer spans 10 Hz again after a look was previewed, where it was squeezed into 20 Hz – 20 kHz.
  • The Room says what it is doing while the Library plays it. Its chip said "Room off" over a room that was playing, and its picture lit every speaker for a stereo song; it now reads the Library player's own room.
  • The FluidEQ Engine installs only the files it was built with. Setup checks every file it puts into the engine's folder against the ones it shipped with and refuses any other, and only the engine's own settings folder can be written to without an administrator.
  • The last English left on screen is in your language, in all ten.
  • The Processes list names every program FluidEQ starts. The helpers that follow Windows' volume, tell FluidEQ which game is in front and play audio another FluidEQ shares with you each have a row saying what they do, and their memory is in the total. The audio capture row no longer says it runs only while you share your audio: a second output and Smart EQ use it too.
  • The Room is built on other people's measurements, and says so. The acknowledgement names them, and SwiftF0's author with them.
  • Every library inside FluidEQ is listed with its licence. The About panel points to a list of all of them, each with its own licence text, beside the notice that says what the main ones are for. A Karaoke Maker project now credits the melody model that actually found its notes, where it named SwiftF0 for notes another model had found.
  • Bloom draws one flower. It no longer draws a second flower under its 3D one, and its petal layers no longer show through each other.
  • The Voice cleaner's model download can be cancelled, and coming back to the page while it downloads picks it up instead of looking stuck.
  • Less memory held for nothing. A graph effect that fails to start no longer leaves its leftovers on the graphics card, and the Karaoke Maker no longer keeps up to thirty minutes of pitch data it never uses.

Most of this release is about what a visualizer somebody else wrote is allowed to do on your machine. A scene is a program, the gallery hands you other people's, and this is the work of going through that as if the next one were written to hurt you: what it may read, what it may reach, how much of your graphics card it may spend, and — the part that matters most — how brightly it is allowed to flash. Nothing about it is visible when it works. The rest of the release is.

Scenes also got faster and better behaved. A new version of one arrives with a line saying what changed, written by the assistant that changed it; a scene is compiled the moment it is added rather than the first time somebody plays it, where Alpine used to stand still for eleven seconds; and a scene no longer stretches over a window it was not drawn for while you are resizing.

On a surround output the speakers the rack skips now stay in time with the ones it runs on — before, a centre or a rear channel could arrive a fifth of a second early. And the rule that only one thing plays at a time turns out to have been about FluidEQ against everything else rather than about the machine: a video started over a playing album left both going.

New

  • A scene from the gallery is held to what it may spend and how it may flash. A visualizer somebody else wrote is a program running on your graphics card, and it is now bounded on every side that matters. How bright and how often it may flash is measured the way the accessibility standard counts it — by the time between one rise and the next, at your screen's real frame rate — which closes a red strobe that read as safe by an older measure and a slow ramp that snapped back. How much of a frame it may cost is a measured ceiling rather than a guessed one, so the scenes that fit stopped being refused. It cannot read a file it names, cannot reach another machine, and never starts while nobody is looking at the screen. None of it applies to a scene you made yourself or one FluidEQ ships: those are drawn exactly as they are.
  • Your assistant is told which part of a scene is the expensive one, and is handed a picture of what it built rather than a door to look through. The Studio's corner says how the scene is doing while you watch it.
  • A scene's content can only change under a higher version. A publication that changes what plays without saying so is refused, by the Studio, by the server and by the app that receives it — so a scene you have cannot be altered underneath you, and the Studio raises the number for you.
  • A new version of a scene says what changed. The update notice and the versions page are both built around one line — "the peaks no longer get cut on wide panels" — and it was optional, so a listener was often asked to take a new version on trust. It is now required when a scene already exists, and it arrives written: the prompt the Studio gives your assistant asks it, every time it changes a scene, for one line about what a listener would notice. It holds two sentences rather than the one an old tweet's worth allowed, it sits across the full width of the publish dialog where the eye already is on its way to Publish, and pressing Publish with it empty turns the field red and puts the cursor in it rather than doing nothing.
  • The Studio says which version is on the bench. The number that decides what a publication is called was readable only on the gallery's own page, so an author tuning a scene could not tell an unpublished version from the released one without leaving the surface they were working on. It sits in the pinned bar beside the project, and takes the app's own accent while it is ahead of what anybody has.
  • The monitors show the visualizer that will be on them. Choosing where a desktop background plays was a choice between two coloured glass tiles. Each monitor now carries a frame of the scene itself, cropped to that monitor's own shape — the one its maker published, or one drawn here from the installed scene — and how it will move is a mark in the corner. The Manage list carries the same frame beside each monitor's line.
  • A scene is ready before anybody plays it. Alpine took eleven seconds to compile, and that was eleven seconds of a still picture the first time it was chosen. A scene is now built the moment it is added or updated, in the background. None of that time was ever about length: the same scene at 51 KB and at 64 KB compiled within 0.3% of each other, and what cost it was one shape in the code that the two heaviest scenes shared.
  • A vote in the Forum can be cast from inside FluidEQ. The vote pill could only fail: GitHub refuses that mutation to every token an app hands out, and has since it shipped. A vote is now the 👍 on the post — the same one github.com shows — so one cast in the app is a vote anybody there can see, and one left on the site lights the pill here.

Fixed

  • A surround output keeps its speakers in time with each other. The channels the rack is not processing were passed straight through while the ones it is processing were held back by its own work, so on a 5.1 or 7.1 output a centre or a rear speaker could arrive up to a fifth of a second ahead of the front pair. Part of this lives inside the engine, so an installed copy will offer an engine update after this version: the timing is right once that is accepted.
  • The Room always receives every channel of the output, and the switch that used to offer a choice the Room overrode now says why it is held instead of pretending the choice is yours.
  • The rack's header fits on one row again. The title, the preset field and its arrows, Reset, the four file actions and the two switches needed more width than the panel has at most window sizes, so the row stood two tall with the actions wrapped underneath and Reset painted over the arrows. The four actions now keep their glyphs and drop their words on a narrow panel, each still naming itself when the pointer rests on it. A switch that something else is holding wears a small padlock in front of the words it belongs to, rather than a sentence on its own line that read as a caption for the power switch beside it.
  • Choosing a rack preset no longer changes which channels the rack runs on. A preset, an imported one and Reset all quietly moved that switch.
  • The live wave no longer freezes while a visualizer keeps moving beside it. It could stop dead and stay stopped for the rest of the sitting, with only minimising and restoring the window bringing it back. Three ways it could do that are closed: one failed frame no longer wedges its drawing for good, it recovers when the graphics card takes its drawing surface away under the load of a scene, and a sound capture that suspends itself restarts rather than handing back the same frame for ever.
  • A scene is not stretched over a window it was not drawn for. While a panel was being resized the last picture was scaled to the new shape, so everything in it leaned until the scene caught up. It now holds still until the box stops moving, and comes back drawn for the size it is.
  • A slow graphics card no longer refuses a scene for the rest of the session. The check that gives up on a scene counted the driver's own warm-up as the scene's cost, and the refusal stuck until the app was restarted — which is what a laptop on its built-in chip, or a machine reached over Remote Desktop, met every time.
  • A listener who asked for less motion gets a slower scene rather than the same one at full speed.
  • One thing plays at a time, whichever two programs they are. Start a video behind a playing album and both kept going: the rule was written as FluidEQ against everything else, so two programs that were both somebody else's were on nobody's register and neither stopped. It could not have noticed in any case — it was told about one program, the one Windows lists first among those playing, which with two of them is as likely to be the one that was already going. It now sees every program that is playing, and a song started here stops all of them rather than whichever was listed first.
  • Share Audio no longer leaves two machines audible at once. A program that launched straight into playing was not counted as somebody pressing play, so the listener's own music kept running beside it. What a machine hears now decides what it asks the other to do, and nothing a description says can silence what is playing at this end.
  • A scene's own page draws its wave where every other surface does. The gallery's page handed each scene the whole panel as the band for its wave, so a scene framed for part of the picture drew it somewhere else entirely: Alpine's aurora crossed the sky on the graph and came down among the mountains on its own page. That page is where somebody decides whether to add a scene, so it is the one place it may not be wrong.
  • A launch that closes itself says so. There were three ways FluidEQ could end without writing a single line anywhere, so a window somebody saw appear and go left nothing behind to explain it. Every launch now says it started, and every refusal to run a second copy says what it saw.

Two large pieces of work. The FluidEQ Engine learns surround: the DSP rack runs on every channel of a 5.1 or 7.1 output, and on top of it comes the Room — a listening room on headphones, rendered through a measured head, with its card on the DSP page, one press to 7.1 where the driver takes it, a listening test that picks the head for your ears, and the same room on the Library's player.

And the visualizers were rebuilt underneath. They now draw on your display's own beat instead of a fixed thirty frames a second, sized by what your graphics card can actually keep up with rather than by a guess — which ends the ghosting that smeared every fast scene, and lets scenes run smoothly on far more machines than before. Scenes also carry the wave their author built them around, so a visualizer now looks the way the person who made it framed it, and the beat they move to is the one in the music.

New

  • The DSP rack on 5.1 and 7.1. Under the FluidEQ Engine the rack used to run on the first two channels of a surround output and leave the rest as they came, so a centre or a rear speaker played without the EQ's effects. Now every channel goes through it: the EQ, the exciter, the compressor and the rest run on each speaker, and the level stages — compressor, maximizer, headroom, the safety limiter — decide once for all of them, so the mix keeps its balance. Stereo-only stages (Dimension, Bass Forge, Bass Punch, Mid/Side) stay on the front pair, the subwoofer feed gets no exciter harmonics, and the surround channels are kept in step with the front. How many channels there are comes from the output Windows plays through, so stereo, 2.1, 5.1 and 7.1 all work without a setting; the DSP page's header has a switch to keep the rack on the front pair only.
  • The Room. A listening room on headphones: every channel of the output becomes a speaker around your head, each through a measured head (MIT KEMAR, at three head sizes) and the room's own early reflections. Stereo music becomes two speakers in front of you, a 5.1 film five and the sub, a 7.1 game the whole ring, decided by what is playing. The Room card on the DSP page is the room from above: drag a speaker to move it, pick a room (studio, living room, cinema, front stage), set its size, walls, distance, centre and sub, choose the head and whether your headphone profile runs after it; a chip beside the switch says what the room is doing this second. The room takes the page beside its controls, measured in metres along its wall and to its speakers; the walls are drawn to the room's size, so a 12 m hall fills the picture with the speakers well inside it and a 2 m booth stands small with the speakers at its walls — Size moves the walls, Distance moves the speakers within them. The speakers the playing stream cannot reach — five of them while stereo plays — are drawn asleep, with a line saying why. A room you shaped can be saved under a name and put back with a press; the head stays yours. Shaping the room is part of Plus; the room and its presets are for everyone. It adds about 11 ms when it runs.
  • One press to 7.1. Games and films send their surround channels only to an output Windows believes has that many speakers. When the room is on and your output is stereo, and its driver takes 7.1, the output panel offers "Set to 7.1" — no administrator prompt, no virtual cable — and Undo puts the output back exactly as it was. A headset whose driver takes stereo only gets no offer; the room is the front stage there.
  • Fit: the head chosen by ear. Heads differ, and the one that puts sounds in front of you is not the same for everyone. The Room card's Fit button (Plus) plays a short sound that walks around you — front, right, behind, left — through two heads at a time, and asks which one sounds more around you. Five pairs, the two strongest heard again with the sides swapped, and the room takes the head that won; Head on the card can still change it any time.
  • Eleven rooms, and yours. The Room's presets sit in the same picker the other stages use, grouped and with each room's numbers under its name: Studio, Living room and Front stage; Home theatre, Cinema, Gaming, Concert hall, Jazz club and Club; Near field and Open air. Every one is a place someone has sat in, with its speakers inside its walls, and every one is run through the engine's room by a test: nothing clips, the walls come back later in a bigger room and not at all in the open air, a louder sub is louder. The What's new tour and the guide have a Room chapter each.
  • Bass management. As a receiver does it: everything under the crossover (80 Hz to start, 40 to 200 on the dial) leaves every speaker and goes to the sub's path, both ears alike, so bass never bounces off the room's walls and a stereo mix keeps its bass in the middle. On by default; Full range puts it back the way it was.
  • Music that fills the room. Stereo music can take the whole ring instead of standing on the front pair: the front pair stays as it is, the centre takes what both sides have in common, and what differs between them reaches the sides a moment later and the rears after that, the way a decoder derives its ambience; the Amount dial sets how much. The chip says "Stereo → whole room" while it runs; films and games are never touched.
  • Every speaker set by itself. Press a speaker in the picture and its panel opens under it: its level, its own distance, its angle as a number, and Mute and Solo — the walk-through a receiver does with a test tone. A speaker moved farther away now arrives later and quieter than the nearest one, the way it would in the room; the Distance dial still moves the whole ring. Solo shuts the other six and leaves the sub, which carries every speaker's bass; a muted speaker wears a stroke in the picture. Mute and Solo are for everyone; the rest shapes the room and is Plus, like the dials.
  • The Room on the Library's player. The Library plays music through its own copy of the rack, and that copy left the room out. Now it runs the same room with the same head, so a song in the Library and the same song from a browser sit in the same room.
  • Visualizers draw on your screen's own beat. They were held at thirty frames a second whatever your monitor could do. Now they draw every frame the display offers — on a 120 Hz or 144 Hz screen, four times the frames of before — with 60 and 30 still there in the View menu under "Frame rate", and sixty held on battery so a laptop is not drained by a picture.
  • A scene is sized to what your graphics card can really do. FluidEQ now times the card itself while it draws. A scene that is too heavy is stepped down to the size that measurably fits and stepped back up once it is comfortable again, instead of being left stretched or stuttering; AMD's FSR and FXAA smooth the result, so a scene running below full size still looks sharp rather than soft. This is what lets the heavier visualizers run on laptop graphics at all.
  • A scene carries the wave it was built around. The waveform's height and where its floor sits are part of the scene now: an author sets them in the Studio and they are published with it, so you see the visualizer standing in the room it was framed in. You can still move the wave for any scene — that is remembered per scene — and a Restore button puts the author's back.
  • The beat is the one in the music. Scenes were driven by a level gate on the bass, which counted a long kick as three beats and a steady bass line as a beat held down — 192 to 287 beats a minute where the music had 120 to 170. It now hears where the music actually puts the beat, reads the treble against what the track really has rather than the whole scale, and lets a chorus carry a scene along and a quiet bar let it settle, instead of inventing a "big moment" on a five-second timer.
  • The View menu and the Studio's column now match. Same order, same group names, each group folding on its own, and the sliders lined up in a wider menu. A Plus visualizer's own controls — the ones its author exposed — can now be set by a listener from the View menu, not only in the Studio.
  • The Graphics card row says which card is drawing. The card in use and the card chosen can differ for a whole session — Remote Desktop can start the app on a laptop's integrated chip and coming back to the real machine does not move it — and nothing said so. The card's own name now sits under the choice.
  • Welcome to Plus, as a moment. Paying now turns Plus on when you come back, and the welcome is a wide card with a scene playing behind it rather than five grey boxes. The Account panel was rebuilt with it: a profile card with your scene as its banner, your membership badge on it, and your links beside your name.
  • The guide shows the Karaoke Maker's own windows. The chapter had one picture of the whole maker and named none of its tools. It now has the tools along the top with eleven of them explained, the lyrics window — the words, which of them are timed, and where the one in hand starts and ends — and the tools that listen for you with the models behind them: what each needs, whether it is on this computer, and what happens to it while nothing is using it. Every picture in the guide was retaken against the interface as it is now.

Fixed

  • The Room no longer sounds boxed into the midrange. The head it renders through was the dummy head's measurement as taken at the ear canal, with that ear's own resonance in every direction of it: nine decibels up at 2.5 kHz, seven down at 100 Hz. Your headphones deliver to your own ear canal, so the resonance was heard twice. The heads are now equalised so that what every direction shares is taken out and only what each direction has of its own is left; the bass comes back with it.
  • Changing output no longer puts a Windows prompt up by itself. When the output you switched to had the engine on it but Windows had not started it there, FluidEQ restarted Windows audio on its own the moment sound was heard — and that restart asks Windows for permission, so a prompt appeared with nobody having asked for anything. Now the card says the engine is not running on that output and asks; its Restart button is the only thing that runs it.
  • Turning a Room dial no longer cuts the sound for an instant. Every change to the rack makes the engine build it afresh and hand the playing state across, and the room was left out of that handover: its new convolvers started empty, which was heard as a tiny hole on every dial move and every dragged speaker. The room now hands its tail over and the new setting fades in over it, like any other stage. The surround channels' alignment lines were left out the same way and are carried over too.
  • Changing outputs no longer puts up an administrator prompt by itself. Under the FluidEQ Engine, switching to an output the engine was not on, or unplugging one, made FluidEQ enable the engine there on its own — a Windows prompt with nobody having asked. Now the output panel says the engine is not on that output, and only its Enable button asks.
  • A launch no longer restarts Windows audio for an engine that is running. The engine writes what it is doing on each output to a small file the app reads, and at launch the app's first read met the engine's first write: the write failed against the reader, the file kept saying what the previous stream had left in it, and the app restarted Windows audio to mend an engine that was working. The write now lands over a file being read, and the engine's log says which step failed and why when one cannot.
  • "The DSP effects are off" now offers the repair that works. The card for an engine that could not start its effects led with "Restart Windows audio", which starts the same engine again and changes nothing — one person pressed it over and over on a fresh install before finding the real fix by hand in Help. It now leads with Update engine, which puts this version's own engine on the machine, and says so outright when the installed one is not the engine this version carries. Restarting Windows audio stays beside it as the quiet second choice. Equalizer APO is gone from that card: it has no DSP effects at all, so switching to it there gave up the very effects the other button can bring back.
  • Fast scenes no longer smear into their own past. The protection that keeps a visualizer from flashing was holding back any quick change in brightness, not just a flashing one — so anything bright moving quickly was blended with the frame before it. A bright bar sweeping the picture was being held on every single frame. It now acts only where the brightness actually reverses, and only fast enough to matter, so fast scenes are sharp while a real flash is still stopped. Two scenes had been slowed down by hand to work around this and can go back to their own pace.
  • Windows playing the music past the engine is now named and repaired. An output has several places an effect can sit and Windows picks a different one depending on the kind of stream, so the engine could be installed, switched on, locked and processing while not one sample went through it — healthy in every reading the app had, and silent to the listener. The engine now counts the audio that actually reaches it. When none does, the card says the music is going past FluidEQ and offers Try another place, which moves it to another of that output's effect slots, at the cost of one Windows permission and a second of silence. In ten languages.
  • Share Audio no longer stops the music it just started. Only one computer plays at a time, and the rule crossed the wire and came back: a pause sent from here returned a moment later looking like somebody over there pressing play, so the music of whoever had just pressed play was stopped. A press is now sent as a press rather than guessed at from a description, and a reconnection or a paused player falling through can no longer be mistaken for one.
  • A scene published by a newer FluidEQ plays instead of vanishing. The number a scene records is what it was written against, not a list of what it needs — but it was read as the second, so a scene from a newer build disappeared from the looks, the gallery and the Studio even when it used nothing this build lacks. Such a scene now plays; one that genuinely needs something newer falls back to the plainer drawing its author chose for exactly that case.
  • FluidEQ says its own version again. In a few places the app named the framework's version instead of its own.

Any account can now make a scene of its own in the Studio, and the Dynamic lighting page lights the devices on your desk while it is open. Most of the rest answers one report — the FluidEQ Engine installed, attached and silent: FluidEQ now works out why on the machine itself, puts right what it can with one Windows prompt, tries every place a sound card's driver may load the engine from, and a bug report says what it found.

New

  • The Studio is open to every account. Without Plus, make one scene of your own: your AI assistant writes it, and it plays on the stage with everything the Studio has — test signals, the code pane, pictures and settings. What Plus adds is shown where you would press it, with a lock that leads to Plus: more projects, keeping a scene in your looks, publishing it, sharing it as a file and putting it on your desktop.
  • Dynamic lighting without Plus. The page shows your own desk and devices, and while it is open your keyboard, mouse and the rest light up with the Studio's starter scene, moving with your music. With Plus they follow any visualizer on the graph, page open or not, each device tuned on its own.
  • Change your name on the board from the Account panel: Change name edits the name and its @handle in place, under the same rules as choosing them.
  • A Studio project can be renamed from its menu, without losing its place on your desktop or in the Plus tab.
  • A new Studio project starts life as a scene, "Hanami at night," instead of a blank one — something to see and hear right away, and a clearer starting point to describe changes from.
  • Opening a folder of scene folders lists every one of them, not only the first. Without Plus the first one opens, and the others wait in the project menu with a lock.
  • Stop all joins Stop in the desktop background dialog, for turning off every monitor's visualizer at once.
  • The user guide says what your PC needs for FluidEQ, the visualizers and the karaoke AI's downloads, in a new chapter under Get started.

Improved

  • Choosing the FluidEQ Engine switches Equalizer APO off by itself, wherever it is registered — not only at the moment you switch, but whenever its own Device Selector puts it back — once the engine is actually on one of your outputs. One Windows prompt, once a session; nothing else on the output is touched, and your own audio vendor's registration (THX, Nahimic, Realtek, or another) is mirrored forward untouched either way. Switching back to Equalizer APO, or uninstalling FluidEQ, puts it back as it was.
  • One Windows prompt for one repair. Two of FluidEQ's automatic repairs could reach the same conclusion seconds apart, asking twice and restarting your audio twice. They now run one at a time, and never twice for the same thing.
  • Ten styles leave the look picker: nine wave forms that drew the same figure as a plainer style, and Canyon. A look you saved on one of them keeps playing, on the style it was closest to, and the now-empty Wave filter is gone with them. 28 free styles remain.
  • Fluid, the titlebar's own drawing on the graph, now matches the titlebar: the same colours under Auto, the same brightness, and the same bar count.
  • A visualizer that fails to draw is offered again the next time, rather than held back until a new version of it arrives. If a scene keeps resetting your graphics driver, avoid it by hand from the picker.
  • Sample scenes say what they are. Their gallery cards read "10-second preview" instead of "free to try": any account can watch them for ten seconds, and Plus plays them in full.
  • The Studio's example ideas are new, and FluidEQ's prompt asks your AI assistant to write a scene's files in place.

Fixed

  • Bug reports say why the engine is attached and silent. A new Outputs section lists what Windows allows on each output, what the setup helper and the engine itself report, and the engine's own reason for passing sound through unprocessed — with the tail of its own log. It also says whether Windows is allowing the engine to load at all, whether the runtime it needs is in place, and whether the engine has ever run on that PC.
  • A bug report carries everything since the previous one, instead of the last hundred and twenty lines of whatever happened most recently: FluidEQ's own log, the engine's, and a new one kept by the engine's setup, which notes every Windows prompt it raised, a declined one included. The engine failures FluidEQ used to pass over in silence are written down too.
  • The engine can be installed, attached and still never loaded by Windows — and FluidEQ now repairs that itself. Several machine-wide things decide whether it may run at all, and a Windows update, a driver's installer, an audio "repair" tool or a disk cleaner can undo any of them long after setup. Until now everything looked healthy and the EQ simply did nothing. FluidEQ now puts them all back in a single pass — the setting that lets the engine load, its registration, the runtime it needs beside it, and the permissions on its own folder — with one Windows prompt, once per run, and restarts the audio so the engine is loaded.
  • Where a sound card's driver never starts the engine, FluidEQ finds the place it will. Windows can load an effect from five different places, and which one a driver actually reads is written down nowhere. When sound goes past an engine that has never run on an output, FluidEQ moves it to the next place, newest to oldest, restarts Windows audio and listens again, and keeps the first place that works — quietly, with no notice while it tries. A driver that reads only the places older versions of Windows used gets the engine there, in place of Windows' own default effect, which comes back if the engine is removed. Your sound card maker's own effects are never replaced.
  • An engine that could not write in its own folder passed every output through in silence. Nothing had ever checked that. The engine loaded, then could neither read its settings nor report how it was doing, which from the outside looked exactly like it never loading at all. It is checked and repaired now, and a bug report says which of the two it was.
  • The card asking to restart Windows audio stops coming back. Returning to the DSP page with music playing showed it again every single time; it is now put away for the rest of the session.
  • Windows audio is no longer restarted with a dialog appearing on its own. It happens quietly, once per run of FluidEQ even if the window reloads, and never where the engine has never run — a restart cannot help there. The card still appears by itself if the restart fails.
  • Where Windows has never once started the engine on an output, FluidEQ says so instead of offering a restart that cannot help. The notice now says what it has already put right by itself, and names what is left to look at — security software, or the sound card's own driver. Meanwhile it offers Equalizer APO, which on such a machine is the only thing there that will process any sound. In all ten languages.
  • Putting Equalizer APO back says when it could not. It used to report success whatever happened; an output it could not restore is now reported, and uninstalling FluidEQ keeps what is needed to restore it.
  • The "Equalizer APO is not enabled for this output" notice no longer flashes up for a few seconds after switching to Equalizer APO.
  • The desktop background button on a visualizer's page is a lock without Plus. It used to open the monitors dialog only to refuse at the end; it now leads to Plus. Its label also fits the panel in German, Russian, French and Portuguese, where it ran past the edge.
  • The EQ and the DSP rack are verified at 44.1, 48, 96 and 192 kHz, so a high-rate DAC gets the same sound as any other output.
  • Windows' own "Audio enhancements" switch is no longer invisible to the app. Turned off for an output, FluidEQ now says so and offers Windows' own Sound page instead of a Restart Windows audio button that cannot help.

FluidEQ now has an audio engine of its own. Your EQ and the whole DSP rack shape everything the computer plays — games, browsers, streaming apps — not only the Library. FluidEQ Plus arrives with visualizers drawn on your graphics card, a Studio to make your own, the desktop visualizer and, in beta, Dynamic Lighting for the RGB devices on your desk.

Highlights

  • The FluidEQ Engine. FluidEQ's own audio engine runs inside Windows' audio service, after your sound card's effects, so the maker's panel keeps working. Setup asks which engine you want — the FluidEQ Engine, recommended, or Equalizer APO — and the actions menu switches between them at any time: one Windows permission and a few seconds of silence while audio restarts, no reboot.
  • The DSP rack for everything you hear. Under the FluidEQ Engine, the Normalizer, Denoise, Exciter, Bass Forge, Equaliser, Bass Punch, Dimension, Maximizer and Master work on every app, with nothing playing in the Library. The Normalizer levels live audio song by song and remembers the songs it has heard, so they are levelled from their first second. Denoise repairs hiss, hum and clicks as they play. The DSP page says where the rack is running.
  • FluidEQ Plus. An optional membership, monthly or yearly, that supports FluidEQ and opens the new Plus tab: Visualizers, the Studio, the Leaderboard and Dynamic Lighting, with the desktop visualizer too. Everything that was free stays free. Sign in from Account in the actions menu, read the terms in plain words, and pay on Buy Me a Coffee in your browser with the same email — the app never sees your card. An account stays signed in on up to five computers.
  • Visualizers. Plus visualizers are living scenes drawn on your graphics card, where the bass, the beat and the treble each move something different. Browse them in Plus → Visualizers, try FluidEQ's free samples for ten seconds with any account, and add the ones you like to play them on the graph. The window can take a visualizer's colours or its light, and some bring birds, petals or stars that drift gently around the app.
  • The desktop visualizer. Put a Plus visualizer behind your desktop icons — a different one on each monitor, moving with the music or calmly on its own. It pauses for full-screen apps, a locked PC and battery power, and comes back the next time FluidEQ starts.
  • Dynamic Lighting (beta). Your keyboard, mouse, mousepad, headset and stand light up with the visualizer on the graph, through Windows Dynamic Lighting and Razer Chroma. Choose Scene, Colour wave, Spectrum or Beat ripple for each visualizer, tune each device, and watch a live preview of your own desk.

New

  • One picker for styles and visualizers. The look's name on the graph opens a searchable panel: 38 free styles filtered by Lines, Fills, Bars, Points, Scenes and Wave on one side, and Plus visualizers by category on the other, each with its picture. Without Plus, the visualizers show locked and explain how to get them.
  • The Studio. Describe a scene to your own AI assistant, paste FluidEQ's prompt, and watch the scene come alive on your music as the files are saved. Projects, a live code pane, test signals, pictures and scene settings share one page, and the preview drags taller. The Bass, Mids and Treble signals play that part of your own music. Publish to the gallery, share as a file, put it on your desktop, or open one of FluidEQ's own scenes to see how it is made.
  • The Leaderboard. Join from the Account panel and Plus members are ranked by hours listened, active days and likes on the scenes they make. One number a day leaves your computer — never what you play.
  • Visualizer versions. Each visualizer shows its version and what changed, your copies update themselves, and the graph tells you when the one you are watching does.
  • A visualizer that fails on your graphics card is set aside. If a scene stops compiling, loses its drawing surface or resets the graphics driver, FluidEQ stops it everywhere it is showing — the graph, the Studio and every monitor behind your desktop — says why, and does not run it again on that computer, so a broken scene cannot take Windows down with it.
  • EQ mode. A menu on the Bands toolbar sets how strongly your EQ and your curves apply (Normal, Studio ×1.5 or ×2), how Q behaves as gain grows, curve smoothing and, under the FluidEQ Engine, minimum or linear phase. Reset puts everything back.
  • Band designs. Save the frequencies and Q of a band layout under a name, and bring it back on any output from the layouts button. Designs never store gains.
  • Switch a band off without losing it. The Active switch and the band's right-click menu take a band out of the chain and keep its settings. Frequency and Gain are dials, and Ctrl-click resets a slider or a dial.
  • Apply a Squiglink import as EQ or as a curve. Apply as curve makes it a headphone correction with its own strength and switch; Apply as EQ asks before replacing the bands you have.
  • The Forum. Help → Forum brings FluidEQ's GitHub Discussions into the app. Anyone can read; sign in with GitHub to post and reply.
  • Playback buttons on the Windows taskbar. Hover FluidEQ's taskbar button for Previous, Play/Pause and Next, even while FluidEQ is minimized.
  • Report a problem privately. Beside the public GitHub issue, Report a problem can send the report to the maker by email: it opens in your mail app, and if no mail app opens, the dialog says so and shows the address.
  • Engine notices that say what to do. An output the engine is not on yet offers Enable, and the output you are listening on is enabled for you. An engine that isn't running offers Restart Windows audio, a newer engine that came with an update offers Update engine, and Remote Desktop audio says plainly that no EQ can reach it.

Improved

  • Styles reworked, and the weak ones retired. Truss is a night bridge with traffic, Skyline a city at night, Terrace a valley with a hopping explorer and Bubbles a storm of glass; Pulse, Sawtooth, Echo, Spikes, Dots and more were redrawn. Nineteen styles left the picker, and a look that used one moves to its closest replacement. Styles crossfade as they change, and the drawing stays on screen when the music stops.
  • Automatic switching. The Auto menu beside the look's name moves to the next look every 10 seconds to 2 minutes.
  • One volume for every player. Library, Karaoke, Online Media and received Share Audio follow the fader at the foot of the window. Karaoke no longer plays at a hidden 80%, and the Media tab's level matches the site's own slider.
  • A safer Clear EQ. It asks first and only sets gains to 0 dB, keeping your bands, Q, EQ mode and preamp. Deleting a band asks on the button itself.
  • Auto normalize guards what you actually hear. Under the FluidEQ Engine it protects the final output with a true-peak limiter instead of a fixed cut, and keeps your own preamp for when you switch it off.
  • Driver type profiles are twelve distinct, gentle curves that start at 50%.
  • Convolution downloads match your output's rate, 44.1 or 48 kHz.
  • The actions menu leads with the audio engine. A card at the top names the engine, says whether it is working and opens its settings — or, when it is not, the way to fix it. Below it the commands are grouped by purpose, each with an icon of its own, and Theme, Animations and Language sit together at the foot. What's new, Report a problem and About live in the Help menu, whose rows have icons too.
  • DSP pages reflow at any window size instead of overlapping.
  • Menus unfold from the control they belong to, with a lit accent edge.
  • Animations are FluidEQ's own choice. Windows' animation effects no longer switch FluidEQ's motion off; set Animations in the actions menu instead.
  • Processes names every part of FluidEQ — the karaoke models, the library scan, the desktop visualizer, Dynamic Lighting, the audio shared with another PC and the FluidEQ Engine running inside Windows itself — and now adds up to an honest total, memory and CPU, measured the way Task Manager measures it.
  • A visualizer you cannot see stops costing anything. The graph, the Studio and a desktop visualizer each let go of their drawing worker while hidden — a minimized window, a covered monitor, a background tab — and pick it back up the moment they are shown again.
  • Up next and the karaoke playlist slide open and closed.
  • Hindi, Japanese and Chinese labels are no longer spaced apart.
  • Signed updates. From this version FluidEQ installs an update only when it carries the maker's signature, and the installed app is locked down against injected code.

Fixed

  • The next song starts while FluidEQ is minimized, instead of waiting for the window to come back.
  • The graph no longer stutters while the automatic preamp is settling. Small, constant adjustments used to redraw the whole chart; now only the curve itself moves.
  • Escape closes the dialog or menu in front and no longer leaves full screen as well.
  • Custom FX stays switched off when FluidEQ reads the configuration.
  • A saved Share Audio code can be edited without the old code coming back.
  • Text boxes no longer change colour under the pointer.
  • When the window has to recover from a fault, FluidEQ's own screen stays up with Reload, Quit and Copy details.

Good to know when you update

  • Your engine stays as it is. Updating keeps Equalizer APO; choose the FluidEQ Engine from the actions menu when you want it.
  • Quitting turns the EQ off. Quitting FluidEQ from the tray takes the EQ and the rack off every output; closing the window keeps them on from the tray. Under the FluidEQ Engine, the sound also returns to normal if FluidEQ is ended or crashes.
  • Looks switch every two minutes until you choose Auto: Off.
  • Motion-sensitive? If you turned off Windows' animation effects for that reason, set Animations to reduced motion in FluidEQ's actions menu.

Meet Fluid, find your way around, and read every detail more comfortably. This release brings an illustrated offline user guide, a dedicated Help menu, a new About screen, and clearer text throughout both themes.

New

  • An illustrated guide inside FluidEQ. Open Help → User guide or press F1 for 16 chapters illustrated with 14 real interface captures. Learn the EQ, headphone corrections, convolution, device profiles, second outputs, Share Audio, online media, Library, DSP, visualizers, Karaoke and Karaoke Maker, with practical steps and troubleshooting tips.
  • Find a chapter and keep your place. Search the guide, jump from its numbered contents, and enlarge any capture. The current chapter highlights as you scroll, and its number stays visible in the contents pane. The guide adapts to smaller windows and works without an internet connection.
  • A dedicated Help menu. User guide, What's new, Fix audio, Report a problem and About FluidEQ are together in the top bar, with keyboard navigation and the F1 shortcut.
  • Meet Fluid in About. A large Fluid mascot joins the FluidEQ wordmark, version, tagline, author credit, and links to the official website and source code. The guide and new About introduction support all ten app languages.
  • A guide you can keep beside the app. The release includes an offline HTML and Markdown guide with its original captures. Open the HTML edition to print it or save it as a PDF.

Improved

  • Clearer labels in Ocean and Black. Descriptions, hints, tabs and ordinary controls use brighter, solid text instead of appearing faded or disabled. Disabled controls retain their separate, quieter treatment.
  • More readable graphs and karaoke captions. DSP scales, meters, timeline labels, lyrics, translations, note labels and score-card captions follow the same readable text colours as the rest of the app.
  • The second-output tour matches your theme. Ocean now shows its own real interface capture, while Black keeps the matching dark screenshot.

Two outputs, each with its own sound. The second output now receives audio before the main output's EQ, and a profile picker sits directly below each enabled device. This update also fixes output-switching crashes and feedback, and keeps Smart EQ from undoing the tuning you chose.

New

  • Choose the second output's profile right where it plays. Its EQ profile picker shows the profiles saved for that device. Switching it changes that output's EQ without making it the main output or replacing the main output's tuning. Each second output keeps its own volume.
  • Game/Video and Music modes for second outputs. Game/Video starts with about 30 ms in reserve and catches up after a stall. Music starts with about 100 ms for smoother playback. The device's own buffering adds to that delay.
  • The welcome tour leads with second-output profiles. The updated screenshot shows the profile picker and both playback modes. Second output, Black theme and Share Audio remain the three NEW features at the top. The tour also introduces the custom graph looks, with instructions in all ten languages.
  • Named profiles comes first in the sidebar, above Automatic profile, so your saved sound is easier to find.

Fixed

  • Headset B no longer inherits headset A's EQ. On Windows, second outputs receive the mix before endpoint effects, so each device applies its own APO profile. Audio received through Share Audio remains part of the local mix.
  • Switching the main output safely stops its mirrors. Changing from A as main and B as second to B as main no longer tries to disconnect an audio connection that has already been removed. Pending starts are cancelled too.
  • Second outputs cannot feed themselves back into the capture. This addresses the repeating sound on the main output and silent second output reported after changing devices. Windows routes mirrors by permanent device identity rather than by matching display names.
  • Smart EQ preserves your tuning while correcting the source. Manual EQ bands and headphone, driver and voicing corrections are no longer treated as faults in the recording. A deliberate cut stays in place while Smart EQ can still correct a real hole or resonance in the source.
  • The welcome screenshot stays inside its frame. Tall captures scale to fit without spilling into the Back and Next buttons.
  • Black theme keeps gray listening bars. Smart EQ's listening columns and progress bars follow the selected theme instead of retaining Ocean's cyan.
  • A failed window can recover without restarting everything. FluidEQ stops playback and reloads its window, with a limit on automatic attempts and a choice to retry or quit if the problem repeats. A packaged app can also restart once after an unexpected main-process error.
  • Karaoke's native model processing runs in a separate process. A model or GPU failure leaves the app running, clears unusable sessions and allows a bounded retry. Reloading the window cancels work from the previous page.

Two big things and a long list of smaller ones. Your other computers can now play through the headset on this one, losslessly and encrypted over your own network. The window has a second face, Black, and it is the default. A welcome window introduces both, then walks through the tabs you may never have opened. Underneath, the whole surface of the app was moved onto one colour ladder, the Library's video became a real player, Smart EQ stopped walking back curves it had already learned, and the DSP rack's presets are protected from clipping at every sample rate.

New

  • Share Audio. A new tab. Choose the computer with your headset as the receiver and it creates a connection code; paste the code into FluidEQ on every other computer and their system audio arrives here, mixed together and shaped by your EQ. Float32 PCM end to end, AES-256-GCM on every packet, private LAN only, and the pairing survives app closes and reboots. Two priorities: Music keeps a larger safety buffer, Game/Video runs with the smaller 100–200 ms buffer to reduce delay for games and video.
  • The receiver shows and controls what the senders play. The playback bar on the headset PC used to say "Nothing playing" while a whole album came in. It now shows the sending computer's song, artist and play state, named after that computer on its third line — a library track, a browser tab, Spotify or a game, whatever the sender's own bar shows. Play/pause, stop, previous/next and the 5-second steps work across the wire, each offered only when the sender's player answers it; pausing keeps the song on the bar so the same button resumes it. Each sender's lane on the Share Audio page carries its own NOW PLAYING or PAUSED line. Smart EQ recognises the shared song, so "save for this song" works while receiving.
  • One player at a time reaches across the wire. With that switch on, starting a song or a browser tab on the receiver pauses the sender, and pressing play on the sender stops the receiver's own players and pauses its browser or Spotify. With several senders, the one that just pressed play is heard and the others are paused. With the switch off nothing is paused, and the bar shows the sender that started most recently.
  • The Black theme. A true black floor, every surface a few values above it, hairlines instead of lightness for the edges, and an ice-blue accent that stays cool beside the rainbow. Every window follows, drawings included: the pitch lane, the Maker's editor and the DSP's scope read the theme per frame. Black is the default for a fresh install and for anyone upgrading; Ocean is one pick away in the menu behind the pulse icon, and the choice is remembered.
  • A welcome window. Two thirds of the screen, one slide per big thing: what this version brought, marked NEW, then Library, the DSP rack, Karaoke, the Karaoke Maker and Online Media, each with a drawing, four highlights and the buttons to press. It opens on every launch until you tick "don't show this again", and comes back once on the next version that has something to show. The menu's "What's new" opens it; the release notes are a link inside it.
  • Native message boxes. The confirmations before reinstalling Equalizer APO or restarting Windows audio, and the notices after, are owned by the window and titled with the product instead of floating free with the page address in the title bar.

Changed

  • One palette. Every page had grown its own surfaces: washes of white at a dozen strengths, black at a dozen more, three tones invented per file. Pane, block, plot and track are stated once and used everywhere; 139 corner radii now sit on a five-step scale; "unavailable" is one dim instead of nine; caution is one amber instead of three.
  • The Library's video is a player. The queue takes a strip beside the picture instead of covering a quarter of it; full screen puts the picture behind the header and the transport, the way Karaoke's stage is; Back closes the picture and keeps the queue; where each video was left is remembered per video, not per session; the toolbar stays over a playing video so the tab has a way back.
  • The window behaves natively. Snap zones, drag-to-restore and double-click-to-maximise work because the window is a real one again; the Black theme takes the desktop blur off it so black stays black.
  • Meters warn at the loopback's own ceiling (-1 dBFS) instead of at full scale: by the time samples rail in the capture, Windows has already squashed the peak.
  • Shift-click selects a range of bands like a file list, on the dots and on the sliders, and the multi-band reset applies to the whole selection.
  • The DSP rack applies to Library playback only, and says so on the page; shared-audio reception and the other tabs are not routed through it.
  • The release-notes dialog is titled "Release notes", since "What's new" is the welcome window now.

Fixed

  • Reception lanes say "Receiving" instead of borrowing the sender's "Transmitting".
  • Background saves preserve complete EQ configurations. Included files finish writing before the configuration that references them, and files from the previous output are removed only after the replacement is ready. Failed saves are reported and can be retried; external profile edits are no longer hidden by a previously saved copy in memory.
  • Updating a profile keeps its name. Update replaces the selected profile; only creating a new profile adds a number when the name is taken.
  • Smart EQ recovers a song's curve instead of walking it back. The loop kept everything it knew about the previous song when the recorder replaced the layer under it, read the song's own curve as a disagreement and stepped it back within one quiet window. Its ranges also vanished from the plot on every bass drop-out and every three seconds of quiet, because a flag flip republished progress with an empty region list. Both fixed, and the fitted-slope modes now converge to the same place from a flat start and a bent one.
  • DSP presets no longer clip or go unstable. The final guard armed only above +10 dBTP, so ordinary preset boosts at unity clipped at the output; it now enforces its -0.1 dBTP ceiling after every processor and holds reduction for a 20 Hz cycle. Filter corners are clamped below Nyquist: at 32 kHz several Exciter profiles had asked for a low-pass above it and got unstable poles. 204 factory cases checked at seven sample rates, every cross-family pair and every local replacement.
  • Neural denoise works at every sample rate, and the Denoise stage draws the noise floor it is subtracting while it measures.
  • Focus rings are never cut off. The halo was drawn outside the control, so any clipping ancestor sliced it; focus now says itself twice, with a hairline just inside the edge that nothing can clip. Quiet buttons, most of the buttons in the app, had shown no focus indicator at all.
  • Stroked icons stay stroked. The DSP reset arrow had a black bowl; the first fix painted Clear EQ and Add band as cyan discs. Both right now.
  • The Videos shelf says which folder it is in, with the same Back and path line the drill-in draws.
  • The folded queue's chip is legible over a picture.
  • Send to Karaoke no longer loses the song to the saved-playlist restore that finished a moment after it arrived.
  • Search fields stop lighting up on hover, and the Online Media search keeps its capsule while the suggestion list is open.

Faster

  • The chain curve is rendered once per chain, not once per frame. With a continuous Smart EQ mode on, the capture rebuilt every layer's response thirty times a second: 2.4 ms a frame with ten filters, 7.8 ms with forty. Now 40 µs and 94 µs.

This patch makes the native DSP engine stay up and stay single, fixes the transport buttons — Stop, Pause, Next and Previous all did the wrong thing in one situation or another — and recalibrates the two bass stages against real music instead of against a test tone.

New

  • Folders look like folders. A folder on the Folders shelf is drawn as a folder holding its cover art, instead of as an album with a different label.

Changed

  • Master's loudness target now allows 9 dB of peak limiting, not 6. Six is a normal amount of work for a delivery and it is the whole of the residual spread for a player: against five sources each peaking within 1.5 dB of full scale, a 20 LU input spread came out at 3.0 LU. Nine reaches the target on the material that needs it most.

Fixed

  • Bass Punch does what its dials say on actual music. Every constant in the stage had been fitted against a kick alone in digital silence, where both envelope followers collapse to the floor between hits and read a ratio no real programme can produce — 28.3 dB alone in silence against 10.0 dB over an ordinary bassline. The top of the attack dial was buying a fraction of what it claimed, and the duck never fully let go. Both are measured against programme material now.
  • Bass Forge's Amount dials reach what the stage can actually make. They stopped at 1 and were multiplied by Mix before reaching the band, and the profiles mix around 0.5 — so the top of each dial delivered half the available effect, and the rest could only be found by opening a second control.
  • A stage monitor stops when you leave its page instead of running on unheard.
  • The DSP no longer sounds different with every filter switched off. The engine's lifetime was tied to whether audio was playing this instant, so every pause tore it down and every play built it again — and each teardown handed the audio back to the fallback player for a moment, so the same track played twice a few milliseconds apart. That is a comb filter, and it took the bass out of the music. The engine now follows whether there is a track to play.
  • Only one audio engine runs at a time. A track change arriving while the engine was still starting was told it had failed, and a failed start is answered by shutting the engine down — so a new one killed the process it had just asked for, over and over, on every change.
  • Switching to Karaoke or Online Media no longer destroys the Library player. It paused the Library, which took the engine down with it and left the player rebuilding itself on the way back.
  • Stop and Pause no longer start the music again. Handing the audio back to the fallback player restarted it, so pressing Stop or Pause silenced the track for a moment and then played it from where it had been. Both now stop, and turning the DSP switch off in the middle of a song still leaves the music playing.
  • Next and Previous work with Repeat one switched on. Both buttons appeared to restart the track from any position: with Repeat one set, the queue refused to move in either direction but still handed back a fresh queue, so the track reloaded and began again. Repeat one loops a track when it ends; it does not disable the skip buttons.
  • Previous restarts the track past ten seconds and steps back before it, reading the position from the engine that is actually playing.
  • The DSP switch keeps the setting you chose. Pausing used to flip it to BYPASSED, rewriting a saved preference to describe a pause. It now stays as you left it and simply goes unavailable, with the header line saying that processing waits for a track.
  • The Auto Headroom preamp stops walking itself down. It corrects far faster than Equalizer APO can show the result, and was counting the same overshoot on every pass while its own correction was still in flight — reaching −11.88 dB where the measurement justified −6, then taking over a minute to give it back. It now only credits itself with what the output has actually shown.
  • Buttons that cannot be pressed now look it. The DSP preset bar's Reset, Save, Export and Import sat in the full accent colour while doing nothing when clicked. Every unavailable button in the app is now drawn as unavailable, and keeps the tooltip that says why.
  • Reset stays on the preset line. In the Denoise header below 1000px, Reset dropped under the picker with most of that row still empty. It now sits beside the picker and wraps only when the two genuinely cannot share the line.
  • A malformed control message can no longer stop the audio engine. The native host checked the size of an incoming command only after making room for it, so a frame that arrived garbled could ask for more memory than the machine had and take the engine down mid-song, with nothing in the log to explain it. Such a frame is now refused by name and the engine keeps playing.
  • DSP visualisers return after the native engine restarts. The replacement process restores the analysis stream before reopening the audio device, so every stage graph resumes without closing and reopening the DSP page.
  • DSP graphs start moving again by themselves. A graph stops drawing when there is nothing to draw, and only moving a control could wake it — so after refreshing the window, changing the output or restarting the audio driver, the spectrum and the stage displays sat still while the music played. They now follow the engine itself and pick up the moment it starts measuring again. The graphs that used to redraw continuously stand down while nothing is playing, so an open DSP page over a silent chain costs nothing, and the time strips clear when the engine lets go instead of joining two moments together as if no time had passed.
  • A clean old process can no longer take its replacement down with it. Every acknowledgement, frame, error and exit is now tied to the native process that produced it. A late code-zero exit during a quick stop/start handover cannot clear the new engine or consume its restart budget.
  • Packaged library scans now keep cover art. Tag reading stays in the background worker, while raw embedded or folder artwork is handed to the Electron host for resizing and caching. Libraries scanned by the affected build automatically re-check missing covers once; existing metadata and cached covers are preserved.
  • Some .m4a files no longer lose their tags, duration and cover. A track whose MP4 structure omits an optional sample table was rejected outright during a scan. Such files are now read normally, and one that genuinely cannot be read is still skipped without interrupting the scan.
  • Leaving full screen now follows the tab you press. Taking the Karaoke stage, the Library or Online Media full screen and then switching workspace returns the window to its normal size, instead of handing the new tab a full-screen layout it was not designed for.

Diagnostics

  • Native engine logs now show the lifecycle that led to a failure: process and generation, start and stop intent, device and analysis state, first analysis frame, restart count, exit signal and the native host's own message. Informational code-zero exits are logged as information rather than errors.
  • Automatic profile writes name their cause. Logs distinguish manual saves, imports, external Equalizer APO edits, convolution migration and the exact edit channel, making repeated Smart EQ writes identifiable instead of showing the same unexplained preset line every few seconds.

FluidEQ 1.6 is about what happens to a song after the equaliser. Version 1.5 could play your library and shape whatever the machine was playing; the shaping was one curve and a preamp. This version adds a full processing rack for the app's own player — denoise, exciter, bass, dynamics, width and mastering, each with a picture of what it is doing — and rebuilds the audio underneath it in C++, in a process of its own, because a browser's audio graph was never going to carry it.

The rack applies to music played inside FluidEQ. It does not change Spotify, YouTube or anything else on the machine — that is what the equaliser and its per-output profiles are for, and they are unchanged. The DSP tab says so at the top of the page.

New

  • A DSP tab, and a chain of nine stages down it. Every stage is off until you turn it on, every stage says what it is doing while it works, and the order is fixed because it is the order the arithmetic makes sense in: Normalizer, Denoise, Exciter, Bass Forge, Equaliser, Bass Punch, Dimension, Maximizer, Master.
  • Normalizer. Measures the complete track once, then applies one stereo-linked gain before anything colours it — True Peak or Loudness, with the measured peak, the integrated loudness and the gain it actually applied all on screen. No pumping and no moving follower, because there is nothing following: the measurement is of the whole file.
  • Denoise: hiss, mains hum, clicks and a neural voice cleaner. Measured from the track rather than guessed at — the scan finds the noise floor and the hum frequency that is really there, and the graph draws the floor the engine is subtracting while it subtracts it. Click repair bridges impulsive damage and leaves anything too long to be a click alone, so percussion survives. Isolate plays only what is being removed, which is the only honest way to judge it.
  • Exciter: three bands, plus Organic and Timing. Each band generates harmonics that were never in the signal — even orders for body, odd for air — over a frequency and a range you set. Organic adds smooth even-harmonic density around a chosen focus. Timing lets highs lead and delays mids and lows for clearer attacks, and adds no harmonics at all. Isolate hears the harmonics on their own.
  • Bass Forge and Bass Punch. Forge synthesises the missing fundamental and reinforces the real one from a single band, for speakers that cannot reach it. Punch works on time rather than frequency — the fifteen milliseconds of attack no filter can get at. Both come with profiles, both can be soloed, and both start off.
  • A fifteen-band parametric equaliser, drawn as the filters actually respond rather than as they were asked to. Peak, both shelves, notch, both passes and band pass; bands added either side of the one you have selected; Character (Focused or Broad) for the way a boost narrows as it grows; Serial or Parallel engines; Minimum or Linear phase; Stereo, Mid only or Sides only; 2× oversampling. Dynamic bands act only when there is something to act on, and the graph says which ones are moving. Forty-seven presets, each bringing its own headroom, and curves imported from Squiglink, AutoEq or Equalizer APO text.
  • Dimension and Maximizer. A stereo widener that works per band and can never change what a mono listener hears — the guard closes on its own when a mix is already out of phase — and a maximizer that raises the overall level without letting peaks past the ceiling, showing how much it is holding down while it does it.
  • Master, with a destination rather than a number. Streaming, Podcast, Audiobook, Broadcast R128 and A/85, Cinema, CD, Vinyl, Club and Reference each set the loudness target and ceiling that delivery actually uses. Auto headroom reduces only the peaks that approach a true-peak ceiling; gain match takes the makeup back off so switching the stage on and off compares the sound instead of the volume; and the last guard — DC blocker, invalid-sample repair, emergency ceiling — reports what it caught rather than silently catching it.
  • Complete racks you can save, name and hand to somebody else. A chain preset keeps every stage, setting and enabled state together; it can be exported as a FluidEQ DSP-chain file and opened again without confusing it with an EQ curve. Per-stage profiles and a crossfade curve you draw and keep sit alongside it.
  • A native audio engine. The whole chain is C++ in a process of its own, named FluidEQ-DSP so a task list says something, and it ends when the app does however the app ended. Two decks with a bounded read-ahead each and a crossfade that moves per sample; MP3, FLAC, Ogg Vorbis and WAV decoded in the engine itself and everything else the library accepts decoded through Windows Media Foundation; a resampler, partitioned convolution, and a loudness meter measuring with the filter the standard actually specifies. It was held bit-identical to the engine it replaced on real music before it became the only rack. If it cannot start, library playback continues unchanged, every DSP stage is visibly disabled and the app says why; there is no second implementation waiting to drift away from it.
  • Smart EQ remembers a song. Listen to a track for two minutes with a correction in place and FluidEQ offers to keep it; the next time that song plays — from the library, from Spotify, from a browser, whoever is playing it — the curve comes back, says which song it recognised, and offers to stop. A shelf per output, the oldest correction leaving when it is full, a star on the row that has one, and Forget reaching the song wherever it was filed.
  • Karaoke songs can hold more than one language of words. A second set of lyrics borrows the first one's clock, so the translation lands on the same notes; the Maker asks which language you are looking at and checks the syllable and note counts against the original; and a Spanish sheet you can see is a Spanish file you can save.
  • Browse the library by genre, and the queue no longer stops when the shelf it came from runs out.
  • Updates install themselves. When one has been downloaded and verified, FluidEQ waits until the window is closed and nothing is playing, then installs it and comes back the way it was. The update that brings you to this version still shows the installer's window — the running app is what asks for a quiet install, so this one is the first that can. Every update after it is invisible.

Changed

  • The title bar carries the app's navigation. The five places sit either side of the live meter — Online Media and EQ on the left, DSP, Library and Karaoke on the right — which gives the workspace its row back. The meter is measured into the true centre of the window rather than being left wherever the names happened to end.
  • The Media tab is called Online Media, and shortens to one word on a narrow window rather than being abbreviated.
  • One typography on all three platforms. Each system's own font, one weight scale, one monospace stack, and figures that do not shove their neighbours around while a value counts up. Two bundled webfonts were tried and taken back out — Segoe UI is sharper at the sizes this interface lives at.
  • The DSP page moves like the rest of the app: the same easings, the same card entrances, menus that open away from the now-playing bar instead of underneath it.
  • An expanded or full-screen response graph follows the thing making the sound. Library art, a Karaoke stage or an Online Media page stays behind the graph even while another tab is selected, and leaving that player returns to one ordinary window state instead of a second competing full-screen mode.
  • Transport controls ask the source that owns them. Online Media uses a loaded page's real Previous, Next, Play and Pause actions when it exposes them; Stop, seek and the Library and Karaoke controls all reach the player whose clock is on screen.
  • The Processes panel explains the split instead of listing Chromium names. Interface, app core, graphics, browser sound, network, pages and the C++ audio engine each say what they do. The engine now reports its own memory and CPU, which Electron cannot measure for a separate executable, and the panel ships in installed builds rather than disappearing outside development.
  • Library DSP now follows Library playback. The native engine starts only while an audio track from the Library is actually playing; while it is idle, the DSP switch stays unavailable and an orange notice says clearly that the rack does not process Spotify, YouTube or other apps.
  • Full-screen transport controls reveal only at the bottom 10 pixels. Once revealed, the bar stays open while the pointer is inside New Look, the Karaoke playlist or the pitch lane, then waits five seconds after the pointer leaves. Those panels can hold an open bar but never reveal one from the middle of the screen.
  • An output without Equalizer APO enabled is marked APO OFF. Its device card explains why FluidEQ cannot change that output and opens APO's Device Selector directly instead of leaving a curve that cannot reach the device.

Fixed

  • Adding a Library folder in an installed build produced zero tracks. The packaged scanner now uses Electron's utility-process port and reads scan requests from the MessageEvent.data envelope that port actually delivers. It also falls back to the in-process scanner if a worker exits before finishing.
  • Stopping the native engine started another one. A deliberate code-zero shutdown crossed the supervisor's exit handler before it had been marked as intentional, so it was reported as a failure and restarted during teardown.
  • New Look's Name field and Karaoke's lower controls no longer move away from the pointer when the bottom transport appears or fades.
  • The rainbow hairline across the top of the header is gone. Rainbow mode remains in the visualisers where it belongs, without adding chrome to the window frame or flashing a horizontal line through the centre when it activates.
  • A library on OneDrive scanned to nothing. Placeholder files were read as empty rather than as files waiting to be fetched.
  • Shuffle moved the playhead. The whole run was shuffled and then searched for the current track, which put it at a random index — it looked like a timer going off, and it was four lines in shuffle().
  • A crackle during playback. The mirror feeding the visualisers was seeking every time a render arrived late; the seek was the sound.
  • Half of an AAC export was silence when an offline render outran its own decoder.
  • A tag block with a bad length no longer costs the file — the bounds check is where the bug is, rather than a retry wrapped around it.
  • Playback enters softly instead of stepping into the middle of a waveform, and a crossfade waits until there is something to fade to. Next and Previous also start the new song at its beginning rather than at the old song's position.
  • The engine follows the output device it is supposed to play to, brings its decks back when the device changes under it, and keeps trying while Windows is between endpoints instead of staying alive and silent after the first failed reopen.
  • The seek bar asks the engine making the sound. It no longer reports a song that has left the screen, follows a stale mirror clock, or carries the last player's Play state into the next source. Moving from Karaoke back to the Library no longer leaves the next track silent.
  • Smart auto-normalize no longer writes the active profile every two seconds while its measurement converges, and it no longer rebuilds every response curve on every analyser frame. Unsaved edits stay unsaved and the renderer keeps its time for the graphs.
  • FluidEQ recognises its own Equalizer APO writes. A config change the app just made no longer comes back through the watcher as an outside preset edit.
  • The Maker's tool popover was measured against the wrong box, so its last row sat under the now-playing bar with no way to scroll to it.
  • A file that had been recovered stopped calling itself an error, and Stop no longer leaves an empty bar behind.

FluidEQ 1.5 adds a music and video library. The app could already shape whatever your PC was playing and open other people's media in the Media tab; it could not see the music sitting on your own drives. Now it can — and because there is finally one thing playing at a time, everything that draws sound got rebuilt around it: one transport at the foot of every tab, a level meter redrawn from scratch, and the title bar's visualiser brought to the response graph.

New

  • A Library tab that reads your folders. Add a folder — or drop one on the tab — and FluidEQ walks it for music and video: MP3, WAV, OGG, FLAC, M4A, Opus, AAC and more, alongside MP4, WebM and the rest. Tags, cover art, durations and bitrates are read into an index that is remembered between runs, and a rescan re-reads only what actually changed.
  • Everything shows up immediately. Files appear the moment the folder is walked, titled from their filenames and grouped by the folder they sit in, dimmed until their details arrive. The scan then fills in the real tags and artwork behind them, so a large library is usable while it is still being read instead of blank until it finishes.
  • Five ways to browse, three ways to look. Album, artist, song, folder or video, each as a dense sortable list, a grid of covers, or a Cover Flow you sweep through with the wheel, a drag or the arrow keys. Folders read either as the tree they actually sit in or as every folder at once, which is the one to use when you would rather find than browse. Click a column to sort by it, click again to reverse, and both choices are remembered.
  • A player that does not stop when you look away. Double-click a song and whatever you were looking at becomes the queue, with shuffle, repeat and a transport at the foot of the window that stays there on every tab. Videos play in the tab with a full-screen control of their own.
  • Cover art without asking anyone. The file's own embedded picture, or a cover.jpg beside it, or a tile drawn from the album's name. Nothing is fetched from the internet and no album name leaves the machine.
  • One transport, for whatever is playing. The bar at the foot of the window follows the tab you are on — the Library, Karaoke, the Media tab — and when none of them is playing it shows what the rest of the computer is: a desktop player, a browser tab. Play, pause, skip and seek reach it from there. Windows only. The bar is always drawn, quiet and empty before anything is chosen, rather than appearing on the first press and taking sixty pixels off the workspace as it arrives — and which player it belongs to survives a restart.
  • One player at a time. Starting a song in the Library stops the Karaoke stage, and starting either stops the Media tab. Optionally it stops players outside FluidEQ too, which is off unless you ask for it.
  • The title bar's visualiser, on the response graph. Fifty-seven forms now draw the live spectrum, including the ten wave forms the title bar has always had, painted by the same code so the two panes finally agree about what they are drawing. Four palettes — signal, rainbow, level and the new Heat, which is the reading rather than a map of it — and ten marks for a lit peak, each with its own behaviour rather than its own silhouette. The Look designer reaches all of it: fill, glow, thickness, how many pieces a form is broken into, and a new Gap that thins a column form without changing its count.
  • A level meter with ten looks. The stereo output meter is drawn in canvas rather than assembled from elements: bar, segments, LEDs, fluid, mercury, needle, pulse, stack, flow and centre.
  • A clip indicator that watches the real ceiling. It reads where the output actually clips rather than where the chain could in theory, and floats clear of the controls beneath it.
  • Auto-normalize gives back the volume it used to waste. It is the same single switch it always was, and off still hands the preamp to you. What changed is what "on" reserves. It used to assume the music has full-scale energy at the exact frequency where your chain peaks, so a 6 dB boost at 10 kHz cost 6 dB of volume — while the music up there is typically 30 dB down. It now measures what is actually coming out and reserves what the programme asks for instead. The arithmetic is still underneath as a floor, so a cold start, a silent room or a measurement that is wholly wrong behaves exactly like the auto-normalize that shipped, and it can never be louder than no chain at all. Measurement lasts the session; nothing about it is stored.
  • The preamp is a dial. A turned metal knob with a cut notch, which reads at a glance in a way a slider on an even range never did.
  • A tray icon. FluidEQ keeps running when the window is closed, so the equalisation does not stop because you tidied your taskbar — see Changed below, because this changes what the close button does. The icon carries a badge when an update is waiting, and its menu will install it, check for one, reopen the window or quit for real.
  • Karaoke files that open elsewhere, and open here. Every format the app says it can read it now genuinely reads, and what the Maker exports opens in other UltraStar players rather than only in this one.

Changed

  • The close button hides the window instead of quitting. FluidEQ carries on in the tray, because a window that is closed is not a reason to stop equalising the machine. Quit from the tray menu when you mean it. Windows shutdown, a session logout and the installer all still end it properly.
  • Only one copy of FluidEQ runs at a time, whichever build it is. Two copies write the same Equalizer APO config and each reads the other's write as an outside edit, so they spend the session undoing one another. The second one now steps aside and says why.
  • Every output gets its own profiles and its own undo. A profile is identified by the output it belongs to as well as its name, so renaming or deleting one on your headphones leaves the identically named one on your speakers alone. Saved copies moved the same way, and both are migrated at startup.
  • The speech model's real cost, in the two sizes it comes in. About 570 MB where your graphics card can run it and about 1.1 GB where it cannot, rather than one number that was right in neither case.
  • Melody detection is no longer offered on the full mix. It reads the isolated voice, which is the only thing it can read honestly, so the button waits for a split instead of returning a plausible-looking wrong answer.
  • The side bar is a card per concept, the tab panels keep their bottom breathing room when they are scrolled, and the graph has its floating toolbar back.

Fixed

  • A format Chromium cannot decode — MKV, AVI, WMA — is listed and marked rather than playing as a black rectangle.
  • A folder on a drive that is not plugged in keeps its songs and dims them. Unplugging a drive never empties a library, and neither does stopping a scan part-way through, nor a file that vanishes while it is being read.
  • An album that does not account for every file in its folder lists the rest underneath it, instead of quietly leaving them out.
  • A file whose tags cannot be read still appears, under its filename, and says that its tags were the problem rather than leaving you to guess.
  • Songs that come with their own lyrics stay out of the library and are counted where they went: they belong to the Karaoke tab.
  • Seeking in the Library player lands where you let go of the thumb, instead of resetting the song to the start. Audio is held as a blob and byte ranges are served for video, which is the whole of what seeking needed.
  • Pressing Play on the album already cued now plays it. After a restart, or after Stop, an album's own Play did nothing while the bar's Play worked.
  • Searching the library searches the library. Typing in the box steps out of whatever folder you are standing in, instead of returning the four matches in that one directory and looking like an empty library. On the folder shelf a search lists every folder holding a match with its path underneath, rather than the single root they all live under.
  • Play in an opened album plays the list in front of you, which is a different record the moment you have typed a filter or sorted a column.
  • The folder chip selects the shelf again. Opening the menu from the chip itself meant the press people make most often cost a second one to dismiss a menu nobody asked for; the arrow beside it is the half that means "which reading".
  • Auto-normalize no longer walks the preamp down to the floor. Two faults, both in the first cut of the measured reserve: the ceiling it held the output under was −3 dBFS, which is below ordinary mastered music and so read "too loud" on every track forever — it is −0.3 now, which is the Windows limiter's own documented figure with two tenths of margin. And the automatic preamp shared its floor with a single band at −20 dB, while two bands an octave apart can peak near +26, so it reserved less than the chain took and clipped by construction. The automatic value reaches −60 dB now; the manual control keeps −20, because it is your thumb.
  • The preamp the config is actually using now reaches the window. Nothing subscribed to the channel carrying it, so the slider and the final curve showed whatever was true when a switch was last clicked and then froze — the config carried −4.36 dB while the sidebar sat at −20.00 dB. The graph no longer computes a second answer of its own and overwrites the real one, which is what made the mode look like it did nothing.
  • The preamp dial paints before it writes, so dragging it no longer waits on a round trip to the config for every pixel, and the number field comes back when auto-normalize is off — a dial is a poor way to ask for −6.5 exactly.
  • EQ Presets stopped the two layers clearing each other. A published headphone correction and an imported set of bands are separate layers, and applying one no longer blanks the other. Both curves are drawn.
  • An output that drops out of the Equalizer APO config now says so. A profile that cannot be read used to take its output out of the chain in complete silence, and the error surfaced later somewhere unrelated. It is written to the log, naming the output and the profile.
  • A profile name the disk cannot keep is refused as you type it. Names carrying : ? * " < > | or a trailing dot used to be accepted and then fail as a file-permissions error that sent people looking at folder permissions.
  • Transcription no longer invents timings it never measured. Whisper sometimes returns a whole verse collapsed onto one instant; those are now discarded rather than laundered into word timings, and the timings it did place are trusted and the rest sung onto the notes. Lines break where the singing pauses rather than mid-phrase.
  • Vibrato no longer shatters a held note into a dozen, and how long a note was held is decided by the stem rather than by counting syllables.
  • A sung aside is no longer read as a section label. [Ooh ooh ooh] and [x2] stay in the song instead of being demoted to a heading and dropped.
  • The graph no longer stacks on itself in full screen, the editor no longer goes see-through when it takes the whole window, and the video player can ask for the whole window and get it.
  • The Look designer opens on the look you are actually using, without repainting the graph the moment it appears and without resuming somebody else's abandoned draft.
  • The release button lets go when it is clicked rather than four seconds later.

Faster

  • The library scan runs in a process of its own, so a folder of ten thousand files no longer holds the window still while it is read.
  • Tags are read through the file's headers instead of loading every file whole — the difference between reading a few kilobytes and reading forty megabytes, once per track.
  • A scan sends its batches rather than the whole library each time. A five-hundred-track folder used to send five hundred copies of a list that grew as it went.
  • One cached thumbnail per cover, not one per track. Two hundred tracks from one album share one file instead of writing the same picture two hundred times.
  • Only the rows and tiles near the viewport are mounted, and the list grows a page at a time, so scrolling a large library stays smooth.
  • Three ways the library held on to things it was finished with are closed, so a long session no longer grows.

FluidEQ 1.4 turns the Karaoke Maker into a complete local production studio. It can split a song into voice and backing tracks, transcribe the words, follow the sung melody and carry the result all the way into the player. The Maker and player were also rebuilt around the same transport, so moving between them now feels like two views of the same song instead of two different tools.

New

  • Separate a song into voice and backing tracks, entirely on your PC. The Karaoke Maker can now prepare both stems from an ordinary song, keep them with the saved draft and restore them when the project comes back. Each stem can be heard alone, and voice, backing and melody guide each have their own level in the shared transport.
  • A stronger local lyric and melody pipeline. Transcription now uses Whisper large-v3-turbo and asks for the song's language instead of guessing; every language the model supports is available. Melody detection uses RMVPE with SwiftF0 as a fallback, follows phrase attacks and singing contours more naturally, and decodes stable notes with a Viterbi pass. Downloaded models, decoded audio and GPU memory are managed explicitly, with a visible way to release the memory when the work is finished.
  • The response graph can live over the Karaoke player. Double-click or use the expanded/fullscreen graph view while Karaoke is playing and the graph is drawn over the stage, with its own visibility and blur controls. Karaoke remains sharp, and the graph menu and pet make room for the song controls.
  • Double-click the Karaoke stage to enter or leave fullscreen. Buttons, sliders and other interactive controls are protected, so an ordinary control click never changes the window mode.
  • The OPRA headphone-correction library replaces the retired AutoEq data. It brings a current catalogue of more than 6,200 products and 12,000 curves, keeps the measurement author and source attached to each correction, and can update independently with the database carried on every version release.

Changed

  • The Karaoke player and Maker now use one transport. Play, seek, skip, backing, voice and melody-guide controls have the same icons, proportions and bottom position in both views. At narrower widths the large level sliders become compact buttons whose menus hold the same settings instead of forcing the bar to overflow.
  • The Karaoke Maker has a responsive, two-level header. Editing tools sit in the header rather than stealing a row below the preview; they stay centred when there is room and wrap onto their own header line when there is not. Playback stays in the bottom transport, and the entire editor contracts for a 13-inch screen without hiding tools or stacking them over the title.
  • Karaoke lyrics focus more like a stage display. The current line is larger and brighter, the surrounding lines fade with distance instead of disappearing into a dark wall, and switching between player and Maker uses a short, low-cost transition without moving the bottom controls.
  • Responsive workspaces use the whole window. Media and Karaoke now fill the available height on tall, narrow windows; the side controls scroll instead of overlapping; the titlebar waveform gives up width before transport buttons do; and graph, player and Maker panes keep bounded heights at the mobile breakpoint.
  • Profiles belong to their output device. Each device now has its own profile folder, deleting a profile asks first, and removing some other profile no longer changes the current selection.
  • Visualizer choices show what they look like before playback starts. The look picker renders a representative frame, and the live-output strip has a compact wave size available from the existing size shortcut.
  • EQ Presets received a layout and interaction polish. Clearing the editable bands no longer makes the selected reference correction lose its name.

Fixed

  • The titlebar waveform now follows the sound the meters are showing. It reads the left and right channel analyzers directly, so opposite-polarity stereo can no longer cancel into a flat trace while both meters are active.
  • The mobile response graph no longer grows the page forever. Its narrow layout has a bounded viewport-relative height and cannot use its own new size as the next flex measurement. Media and Karaoke no longer stop halfway down the same narrow window either.
  • Karaoke controls no longer collide in expanded graph or fullscreen views. The graph options, titlebar, pet, chords, lyric-size control, playlist button and bottom transport all reserve the space they need, whether the top bar is shown or hidden.
  • The Karaoke transport no longer drifts between player and Maker. Both surfaces use the same bottom inset and padding, percentage labels no longer squeeze into neighbouring controls, and the backing-only label remains inside its own menu at small widths.
  • The Maker's header and tools no longer overlap. Long titles truncate, action groups shrink before they collide, and tool buttons keep a consistent icon size instead of expanding with the row.
  • Core tab changes no longer flash through a dark frame. EQ, EQ Presets, Voicing and Convolution remain opaque while their content changes, matching the already-stable Media and Karaoke tabs.
  • Saved Karaoke stems and guide vocals recover cleanly. Restored stems are adopted instead of separated again, the guide vocal can rebuild itself after an interrupted render, solos stay contained, and drafts follow the source content rather than a temporary filename.
  • The embedded Media browser and app window have an explicit content security policy. External navigation is constrained to the intended browser flow, and Chromium page zoom cannot accidentally resize the whole interface while the timeline keeps its own wheel gestures.
  • Output and profile state no longer invents folders for missing devices. Windows output discovery is kept separate from profile writes, and an absent output cannot leave behind a phantom profile location.

Mostly the Auto normalize switch, which was getting the level wrong in three different ways and showing the wrong number in a fourth. Also: the Karaoke Maker had no way back to the song you started from, and these release notes had no obvious way out.

New

  • Restore original, in the Karaoke Maker. Next to Undo and Redo. It throws away the whole session and rebuilds the karaoke exactly as it was imported, saved draft included — for when an edit went wrong several steps ago and undoing one at a time is not the way back. It is itself undoable, so pressing it by mistake costs nothing.

Changed

  • Turning Auto normalize off no longer puts back an older preamp. It used to restore whatever the slider was last moved to, which meant switching off could jump the level to a number from some earlier session. Nothing is remembered now: switching off keeps the level Auto normalize had arrived at and hands the slider control of it, so the switch decides who is in charge of the volume and never the volume itself.
  • The release notes have an OK button. The corner ✕ is what you reach for to escape something, not what you press when you have finished reading. OK sits at the bottom right, outside the part that scrolls, so a long changelog never hides it — and it holds the focus, so Enter closes the notes.

Fixed

  • Turning Auto normalize on shows the reserve it just worked out. On the Karaoke tab the preamp stayed on the old manual number, so the switch looked like it had done nothing. The sound was always right; the display was not, and only on tabs without the response graph — everywhere else the graph quietly corrected it a moment later.
  • Turning Auto normalize off no longer makes everything louder. It published 0 dB instead of the reserve that was actually in force, so a chain holding back 11 dB got 11 dB louder from one click of the switch whose whole job is to stop it clipping.
  • Preamp: -19 is read as -19. A preamp line without the unit was read as 0 and still reported as a successful import, so the file loaded and played at the wrong level with nothing to say so. The unit is optional now, like every other tolerant field in that parser.
  • An imported preamp is no longer overruled without a word. Automatic normalization recomputed over the value the file asked for, so what played was our number rather than the exported one. A file that carries a preamp line now switches automatic mode off, and the import result says that it did.
  • Profile names can no longer reach outside the profiles folder. A saved profile is stored under the name you give it, and that name was being used to build the file path without checking it stayed where it belonged. Loading, saving, renaming and deleting all check now, and a name that would land somewhere else is refused rather than quietly rewritten.
  • Links only open in a browser if they are browser links. Anything the app offered to open externally was handed to Windows as-is, which is not the same thing as opening a web page. Only http and https are passed on — the Remote Media player already worked this way, and now the rest of the app does too.
  • Rainbow mode switches off properly the first time you win it. Turning it off in the session it was earned in left most of the colour on screen — parts of the app went back to normal and the rest kept celebrating. It looked fixed after a restart, which is why it was easy to miss. Off means off now, whenever you ask for it.

1.2.0 could play a karaoke file. This one can make you one, out of a song and nothing else: the words come out of the audio, the timing comes from you tapping along, and the melody is read off the recording. What comes out is a file the Karaoke tab plays like any other.

The rest is about trust. A copy of FluidEQ can now tell you where it came from, and updates arrive only from the place the build in front of you was actually published.

New

  • A Karaoke Maker. Open a song and it walks the whole way to a finished karaoke file. The lyrics can be transcribed from the audio, or pasted and kept. Timing is recorded by ear — play the track, mark each line as it starts and ends, then go back and mark individual words where a line needs the detail; every mark can be nudged earlier or later rather than redone. The melody is analysed off the recording into notes you can hear, split, delete, retune or add to by hand. There is a count-in preview, undo and redo throughout, and the result goes straight to the player or saves as a project to come back to.
  • The lyric transcription downloads a model the first time you use it. It is a speech-recognition model fetched from Hugging Face, and it is the one thing in FluidEQ that reaches the network for something other than an update. It happens on your press, not on launch, and every other part of the Maker — the timing, the melody analysis, the audio itself — runs on your machine. If you would rather nothing was fetched, paste the words in instead and the rest of the Maker works exactly the same.
  • A copy of FluidEQ will tell you where it should have come from. The licence lets anybody copy, change, rebuild and sell this program, which is as it should be — but it also means somebody can hand you a build and call it ours. There is now a panel naming the official site and the official repository, and saying plainly that a download claiming to be official without a valid Windows signature is neither.

Changed

  • Updates come only from where your build was published. The build from GitHub updates from GitHub, exactly as it always has. The signed build updates only from its own address and never from the public repository, because signed installers are not published there. Which one a copy is was decided when it was built and is compiled into it, so it cannot be changed by editing a file in an installed copy. Each side checks what it downloaded before running it — the signed one that the signature is ours, the GitHub one that there is no signature at all — so an installer that ended up in the wrong place is refused rather than installed. Development builds do not update at all.
  • The window opens at nine tenths of your screen, centred, or maximised on anything below 2K. It used to open at a fixed 1428×625, centre itself for that height and then grow to 1036 from the same corner — which, on a 1080p screen, put its bottom edge under the taskbar on the very first launch.

Fixed

  • Signing in works in the Media tab. Sites that hand their sign-in to somebody else — Suno through Microsoft, Twitch through Amazon, and the Apple, Discord and Facebook buttons wherever they appear — used to reach a page the player refused, which looked like a broken login. Each of those sign-in hosts is now named, one at a time and never as a wildcard, because a page here can hold a live login and the list is the whole boundary. Google is listed too, but decides for itself whether to allow an embedded sign-in and often will not; that one is not ours to fix.
  • Suno's old address resolves. suno.ai still answers and redirects to suno.com, and a redirect the player would not follow was a dead link.

FluidEQ has a Karaoke tab: songs you already own, with the words, your pitch drawn against the melody, and the chords guessed out of the backing track. It ships no music and downloads none.

The rest of the release is about the config on disk. Your own APO commands are a layer you can see, an edit made in APO itself no longer goes unnoticed, and auto-normalise takes back a decision from 1.0.0 that you will hear.

New

  • A Karaoke workspace, built out of your own files. Open a song, add a folder, or drop files anywhere on the window. Audio can be MP3, WAV, OGG, FLAC or M4A; lyrics can be .lrc for line timing, enhanced .lrc for word timing, or an UltraStar .txt for syllables and real target notes. Files sharing a name pair into a playlist you can reorder, remove and pick through, and where the pairing is genuinely ambiguous FluidEQ asks instead of guessing. A file this build of Chromium cannot decode says so rather than being called supported.
  • Lyrics that follow the track rather than a timer. Every frame asks the audio where it has got to instead of counting forward on its own, which is what keeps a long song from drifting away from its words. Where the file carries word or syllable timing, the current line fills as it is sung. Scroll to read ahead without moving the music, click a line to jump to it, and set the text size to whatever you can read from across the room.
  • A pitch lane, honest about where the target comes from. Turn on a microphone — chosen from a list, never opened without your press — and your pitch is drawn live, named as a note, and marked high, in tune or low. Target notes appear only when the song file actually contains them, which today means UltraStar. An .lrc or a bare audio file gets a working live tuner and a line saying there is no target pitch in this song, not an invented melody drawn under your voice. There is no score in either case.
  • A melody guide tone. Where a song does carry target notes, FluidEQ will play them as a soft tone at its own volume, so the melody is something you can hear rather than something you read off a lane.
  • A performance review. The places you sang high, sang low, or missed are collected as a list of times; pick one and the song jumps there and counts you in — 1, 2, 3, GO.
  • A chord guide. FluidEQ analyses the backing track and shows the chord under the playhead and the one coming next, with the seconds until it arrives. It is an estimate made from audio and says so: it reports its own confidence, admits when it cannot find a stable chord, and knows major and minor triads and nothing else.
  • A full-screen stage. The window goes to the lyrics, the floating controls fade after two seconds of stillness and come back the moment you move, and the FluidEQ header can be hidden. Playlist, microphone panel and pitch lane all resize and remember their sizes. Your playlist, its order, the selected song and the playhead all come back when FluidEQ next opens.
  • The microphone is measured, not recorded. It opens on an explicit press and is released the moment you leave the Karaoke tab. Nothing is recorded, nothing is sent anywhere, and it is not played back through your speakers, so you hear yourself in the room. Echo cancellation, noise suppression and automatic gain are all turned off, because each of them rewrites the pitch being measured — which also means a microphone sitting beside a loudspeaker will be heard singing along with the record. The song keeps playing when you visit another tab. The microphone does not.
  • What it is not. Karaoke does not strip the vocal out of a recording, change its key or its tempo, or separate it into stems — an instrumental is something you bring. There is no permanent library and no account.
  • Your own Equalizer APO file is a layer like any other. 1.1.0 gave each output a file FluidEQ creates and never writes to, for the APO commands that have no interface here — and then ignored what you put in it. Its filters are now drawn on the graph in orange, named by a chip that switches off like every other layer, and counted in the preamp, so a custom file that boosts no longer pushes the chain past the reserve. Only Preamp, Filter and GraphicEQ are read: a Plugin, Copy or Delay line gets the chip but no curve and no headroom, and the chip's clear button empties the whole file including those lines.
  • An edit made in Equalizer APO itself shows up while FluidEQ is open. Change a gain by hand in one of the layer files and the app adopts it, redraws it, re-derives the preamp and saves it into your profile, with the layer's chip renaming itself to say where the change came from. Where the edit uses a filter type FluidEQ cannot represent it is refused silently — your file is left as you wrote it, but nothing on screen says the two have stopped agreeing.
  • The AutoEQ and convolution searches remember what you looked for, the way the player's search box already did, with one press to forget all of it.
  • The warranty disclaimer is something you can read, in your own language. Sections 15 and 16 of the GPL have always been in LICENSE, in every file header and on the installer's licence page — in all three in the register of a licence rather than of a sentence, which is a way of being present without being read. The same terms are now put in front of you once, on first run, in all ten languages: no warranty, and no liability for damage to hearing, to equipment, or to data. Nothing in it is a new term. It says that some countries do not allow those exclusions and that where that is so, the law wins and the notice takes away none of the rights it gives you. There is a Quit button beside Accept, because a notice you can only agree to is not a notice, and the same text stays in the About panel to be re-read. What is stored is the version of the notice, the version of the app and a timestamp — anything FluidEQ cannot read back is shown again rather than assumed accepted.
  • A release can say that it must not keep running. If a version ever ships a fault bad enough that carrying on with it is the wrong thing, the release that fixes it can be marked as such, and installs from 1.2.0 onward will say so rather than waiting to be noticed. It asks rather than blocks: the notice closes, says plainly that closing it means later and not no, and returns every fifteen minutes until the update is in. The download runs while it sits there, and if the download or the install fails it says which of the two and gives you the release page as a link and as text. It is not a kill switch — Equalizer APO applies its files whether FluidEQ is running or not, so nothing here can take your audio away.

Changed

  • Auto-normalise reserves against the whole curve again. 1.0.0 started discounting a boost by how little music usually holds at that frequency — up to 8 dB of it in the treble. "Usually" is not what a preamp is for: a recording that does reach full scale at 12 kHz was clipped by the reserve meant to prevent exactly that. The allowance is gone and the reserve is measured against the real peak of the chain. A chain whose largest boost is in the treble is several decibels quieter than it was in 1.1.0, everything else is 0.2 dB quieter, and nothing gets louder — some system volumes will want turning up. A cut no longer buys makeup gain unless the whole curve sits below unity.
  • A convolution file is measured rather than sketched. The impulse response is analysed when it arrives, so the graph draws the file's own response instead of the approximation published beside it, and a locally imported WAV — which used to draw nothing at all — has a curve. The preamp reserves against what the file actually does, so most published responses come out slightly louder and a hot one is brought down rather than left to clip.
  • Squiglink is a link now, not a second library. Browsing GadgetryTech from inside FluidEQ has gone, and the measurement-source picker with it. In its place is a guided import: open Squiglink in your browser, export the EQ text, then paste it or open the file here, with the curve previewed before anything is applied. This is the smaller feature and it is worth saying so — what arrives becomes your editable bands rather than a headphone layer of its own, so it does not survive clearing the EQ. A correction already applied from that database keeps playing; the pickers no longer name it.
  • The Ctrl+W cycle trades a stop nobody used for one worth having. Still five stops: Everything → Layers over wave → Curves only → Clean → Wave only. Layers only has gone — it differed from Curves only in the weight of exactly one line — and Clean takes its place: every curve, the grid, the wave and the band handles still drawn, without the nine tinted columns that brighten as Smart EQ hears each range. That is the state for watching a measurement run rather than watching a wash of grey settle over the curves.
  • Every curve can be switched off from the View menu, not only the EQ line — and hiding that line now only hides it, leaving the other layers up to read against, where before it quietly rearranged the whole plot. The legend runs in the order Equalizer APO applies the layers. The graph also remembers per tab whether it was open and how the space was split, so closing it on Convolution leaves it open on EQ.
  • The Video tab is called Media, in all ten languages, and Config sits at the far edge of the tab strip, away from the tabs that change the sound.
  • Euphoria mode is called Rainbow mode.
  • What's new opens on the version you just installed. After an update the dialog is answering "what changed", so it shows that one section and stops. Opened from the actions menu it is answering "what happened", and the whole history is there as before.
  • The installer offers Hindi. The app has shipped ten languages and the setup offered nine, and the setup is the one place a missing language cannot be put right afterwards: it picks its language before the app has ever run, so somebody handed English there never reaches the menu that would have told them otherwise.
  • The Support panel says what is actually sold. It told you that nothing was behind a paywall, which stopped being true the moment a signed build went on sale. It now says the arrangement as it stands: the source is public and you can always build it yourself for nothing, and what is sold is the signed, ready-to-run build. Nothing is tracked, and that has not changed.

Fixed

  • Smart EQ at zero strength keeps its chip. A measured correction turned all the way down was dropped out of the state altogether, which took away the slider that would have brought it back.
  • A dropdown you can search puts the cursor in its search box. The list is drawn outside the control it belongs to, and the field was being looked for inside it, so opening one and typing went nowhere.
  • The release notes fill the width they are given. Every line of this file was drawn as a paragraph of its own, so its eighty-column wrapping became the layout: the prose broke two thirds of the way across a wide window, and the gap between two paragraphs looked no different from the gap between two lines of one. Lines join until the next blank one now, the way list items always did.
  • A button near the top of a dialog takes a press. The titlebar's drag region runs the full width of the window, and a dialog painted over it changes what is drawn rather than what the mouse hits — so About, Support, the bug report, the audio troubleshooter and the score card all had a dead band across the top, their close buttons included.
  • The band sliders ask for a length the card can give them. With the graph switched off the track was measured against 330px of surrounding chrome where there are really 544px, so every window between the 620px minimum and roughly 950px promised the row about 214px more than it had: the panel scrolled, the captions bunched at the top and the lower arrows went off the bottom edge. Above 950px the arithmetic came right on its own, which is how it survived.

Equalizer APO's configuration used to be one block per output, with every layer poured into a single run of numbered filter lines and nothing in it saying where the voicing stopped and your own bands began. It is now a file per layer, and that is where most of what follows comes from: a layer with a file of its own can be switched off without being taken apart, told apart from your own tuning when FluidEQ starts up again, read on disk, and sent to somebody else.

New

  • Any layer switches off without being lost, and every correction has a strength. Switching one off leaves its file out of the chain rather than clearing its settings and stashing them, so pressing the switch again puts exactly the same thing back, and it survives a restart. Your bands can finally have a switch of their own, so can the impulse response, and Smart EQ has joined the driver, headphone and voicing layers in having a strength — half of a measured correction is a reasonable thing to want when the measurement is more confident than you are.
  • A headphone correction is a layer of its own, on a page of its own. Applying a published measurement used to replace your bands outright, so clearing the EQ threw the correction away with your tuning and Smart EQ read it as error and flattened it over a few passes. It now sits beside your bands with its own chip, strength, switch and curve, from either headphone database.
  • Continuous EQ. Pressing Smart EQ applies one measurement whole; this is a mode instead — measure, move a fraction of the way, measure again, for as long as there is music — so one track's emphasis and the next one's cancel out and only what every record agrees about survives, which is your headphones and your room. Detail corrects peaks and dips, Balance also evens out a bright or warm recording, and Target brings every record to the same tonal balance.
  • A Config tab, showing what is actually on disk. Every other panel shows what FluidEQ intends; this one shows what Equalizer APO has got, and the two differ exactly when it matters — after a hand edit, another tool, a restore from backup. Each output also gets one file FluidEQ creates empty and never writes again, applied last, for the APO commands that have no interface here.
  • Send a chain to somebody. A whole output's tuning exports to a .fluideq file and imports back, always onto the output you are listening to, because that is the only one you can judge the result on.
  • Media buttons in the title bar — previous, play/pause and next, commanding whatever is playing anywhere on the computer rather than only our own player. Expect up to a second between the press and the track changing. Windows only.
  • Play to a second output at the same time, with a level for each and nothing to install. What it is not: the sound arrives about a fifth of a second late, so it suits music in another room and not video; it runs only while FluidEQ is open; and every mirror carries the correction of the device you are listening on, because that is in the sound before FluidEQ sees it.
  • A level meter that reads in real decibels, both channels separately. Nothing on screen answered "how loud is this, really" before.
  • Eight genre voicings — rock, metal, pop, hip-hop and R&B, electronic, jazz, classical, acoustic — beside the five that were already there.
  • The graph remembers each of its sizes separately, so arranging the plot for a video in full screen no longer rearranges it for editing bands afterwards. The legend is a set of switches, one per curve, and the View menu can switch off the level meter and the title-bar wave.
  • The player stays signed in between runs. It ran on a session thrown away when the app closed, which made it no use to anyone with an account: a signed-out music service shows ads to somebody who pays not to see them. The guarantee changes from "nothing is kept" to "nothing is kept that you did not ask for, and one press throws all of it away" — that press is a new button in the player's toolbar, always visible.

Changed

  • Auto-normalise gives volume back as well as taking it away. The preamp could only ever attenuate, so a chain that merely cut made everything quieter and nothing put it back. It will not invent volume either: a narrow cut restores nothing, because away from its centre the chain never moved.
  • Smart EQ measures the record, not what you have already done to it. It treated everything it heard as correctable, so it could not see past a cut — a range dropped 20 dB by mistake hid the very evidence that the cut was wrong.
  • Loudness has been removed. It was an EQ curve wearing the name of something Equalizer APO cannot do: a loudness control on an amplifier raises perceived level, and APO has filters and nothing else. Measured, it lifted the ends by 4.5 dB, the preamp reserved 4.1 dB against that boost, and the midrange ended five decibels down — turning it on made things quieter. The equal-loudness voicing curve stays, being the same idea stated honestly.
  • Spotify and SoundCloud have gone, and Suno is being tried. A button leading to a page that does not play is worse than no button. Spotify signed in, browsed and searched perfectly and never played a note, its audio being encrypted with a system this runtime has no decoder for; SoundCloud refused a sign-in by all four of its routes without producing a single error anywhere we could see. Suno is on the same terms as everything else: it stays only if it works.
  • A published correction reaches Equalizer APO as the curve it was published as. Some ship as a list of points rather than as filters and APO draws those natively, where until now it was FluidEQ's approximation being written out. Nothing on screen ever showed the difference — only the file APO reads.
  • On startup, FluidEQ can tell which layer wrote what. It used to refuse to read a config back at all while any layer was live, because a voicing and a row of hand-placed bands looked identical in the text. That refusal now applies only to a flat config, and which layers you had switched off comes back too.
  • Zoom moves in half steps, about 9.5% a press instead of 20%, because settling on a size is what people actually do with it.

Fixed

  • A chain somebody sends you can no longer carry code into Windows audio. A .fluideq is advertised as a settings file, and the part of it holding your own Equalizer APO commands was written into the config word for word — into a file included all the way up to APO's own, in a language that can load libraries. An imported block is now refused if any line loads code, includes another file, or carries a control character that could end one command and start another; the rest of the chain still imports, and the panel says what was dropped and why.
  • The player no longer writes sign-in addresses to the log. Every address it navigated to was recorded whole, and the report-a-problem tool attaches the tail of that log and offers to post it publicly. The tail of a sign-in address is not a location, it is the credential.
  • The player stopped handing a storage permission to anything that asked, and an impulse response file name can no longer smuggle a second command past the config writer — a newline in a name ends a line in an Equalizer APO config.
  • A band drag starts where you grabbed it. Pressing a handle threw the band upward before the cursor had moved a pixel, and grabbing one off-centre snapped it by however far off you were: it was measured against the drawn curve rather than against your pointer, and the two had stopped being the same curve.
  • A layer that is switched off stays switched off. Smart EQ preserved the voicing, the driver and the headphone correction whether or not you had bypassed them, so it looked for shapes that were not in the sound and rebuilt the layer you had just switched off inside its own.
  • Importing a chain no longer destroys the thing being imported. It dropped the headphone correction on the way in, then saved the profile back from what was live a moment later.
  • Clear on the AutoEQ page clears the correction and leaves your bands alone. It flattened every band, which was right back when clearing a reference meant undoing the bands it had written — so it wiped a tuning it no longer owned and left the correction playing with nothing on screen naming it.
  • A strength slider can always be moved. It was disabled while its layer was off, which pinned the strength wherever it happened to be. Dragging it brings the layer back, dragging it to zero switches the layer off, and switching a layer on from zero goes to full.
  • The bands do not vanish when every one of them is at zero. A flat chain is written as a preamp and no filter lines, because a band at 0 dB does nothing — but read back, "no filters" and "no bands" look identical.
  • A switched-off equaliser cannot be edited, and neither can the band handles with the engine off. Both stayed draggable while looking inert, so you could spend a minute shaping a curve connected to nothing.
  • The EQ chip is about the bands, which is all it holds now; it was naming itself after the headphone model, so one pair of headphones appeared twice in the same row. The headphone chip no longer vanishes at zero strength either.
  • Opening a dropdown no longer takes the pane's scrollbar with it, freezing the panel and letting its content jump sideways into the gutter. The View menu measures itself rather than running off the bottom of the screen, and the expanded graph has its left edge back.

The version number is the only thing here that changed suddenly. FluidEQ has been a finished program for a while — automatic per-output profiles, four processing layers written into one Equalizer APO config, a measured correction, ten languages, an updater that works. What it did not have was a video playing underneath the spectrum, and that is what the last two releases were for. It works now, on every site it offers, in every size the graph comes in.

So: 1.0. Not because a milestone was reached, but because calling it 0.x had stopped being honest.

New

  • Switch any layer off without losing it. Press a chip in Also applied and that layer leaves the config — the chip stays, dimmed, and pressing it again puts the same settings back. Nothing is recomputed, because nothing was lost. It is how you find out whether a correction is actually an improvement: the same passage, both ways, a second apart. Voicing, driver, Smart EQ and loudness for now.
  • The bands have a chip of their own, named by the model they came from and the measurement behind it, or by how many bands there are when you placed them yourself. It says (modified) once you have moved them, so the label never claims a curve that is no longer on screen.
  • Loudness moved in with the other layers, off the waveform meter. It is written into the Equalizer APO chain like everything else in that row, and it belongs where the rest of them are. Euphoria mode now brings it with it, and takes it away again.
  • Full screen, rebuilt. The window takes the whole display, the side panes go, and the picture reaches all four corners. The controls become a small card that fades out when nothing is happening and comes back when you move — click the drawing to hide or show them at once. The creature stays in the corner.
  • Keep the top bar in full screen, from the View menu, if you would rather see where you are. On by default.
  • See through and Blur are in the View menu, where there is room for them, instead of taking half the control strip.
  • Every site remembers where you left it. Switching between YouTube, YouTube Music, SoundCloud, Bandcamp and Twitch now comes back to the track that was playing, a few seconds before where you stopped, and carries on. Only one player ever runs, so leaving a site still stops it — this is what makes coming back cost nothing. Pressing the site you are already on goes to its front page.
  • The wave can hang, mirror, or grow from the middle. Ctrl+I steps through four shapes: standing up, hanging down, mirrored into two halves that share the height, and mirrored outward from the centre line.
  • Stretch the wave across the whole card. With the grid hidden there is no scale to leave room for, so the drawing takes the space the labels were using.

Changed

  • Less is taken from your volume. The preamp reserves headroom against what music actually contains rather than against a flat signal to 20 kHz — a boost at 12 kHz can never cost its full value, because nothing arrives at full scale up there. Chains that were quiet for no audible reason get several decibels back, and are no closer to clipping than before.

  • Smart EQ always measures from flat. The switch that asked has gone: a region already cut hard has almost no energy left in it, so measuring the corrected output means the correction hides the very problem it is causing. There was only ever one right answer.

  • The chain reads in the order it makes sense. Convolution, then the driver correction, then your bands, then voicing, then Smart EQ last — physical, intended, taste, measured. Nothing sounds different; the config is now readable at two in the morning.

  • Only the live trace gets the rainbow. The other curves keep the colours their legend names them by, which is what makes the legend worth having.

  • Vimeo has been removed. It renders its listings entirely on the client, from an API that returns nothing to a session it does not recognise — search, Staff Picks and the watch feed all arrive with their tabs and filters and not one result. FluidEQ's player is signed out by design and always will be, so that is the only Vimeo it can ever see. A button onto an empty page is worse than no button.

  • The player signs into nothing, and remembers nothing. Its session is held in memory and dies with the app, and sign-in is turned away at each site's own front door. No account of yours is ever the identity behind what it does.

  • A refused link leaves you where you are. Following one that goes outside the player used to move you to a different site's front page; now the page you were on simply stays, with a note saying what was refused.

  • The spectrum is drawn on a canvas. Both of them — the one across the graph and the small meter in the title bar. They were built out of SVG paths before, which meant that thirty times a second the browser was handed a fresh description of the shape as text, several thousand characters of it for the denser looks, and asked to re-read it, re-measure it and repaint that part of the window. A canvas is told to draw and draws. Nothing about the looks changed: the same forty-six of them, the same palettes, the same glow, all built by the same code — only what receives the drawing is different.

  • The graph no longer redraws itself for sound it is not showing. The chart woke up around twenty-two times a second whether or not the wave was on screen, rebuilt everything it owns, and threw the result away. Now only the parts that actually show live audio listen for it: the wave, the clipping warning, and the measurement overlay.

  • The controls above the graph sit on a panel of their own instead of floating loose over the drawing, and in full screen there is one panel rather than a panel inside a panel.

  • The filter types are drawn against a shared 0 dB line, and named. They used to sit on two different baselines — a peak rising off the floor next to a low shelf falling from the ceiling — which made a shelf read as a treble cut and made the pairs look swapped. They are mirror images of each other now, and the band editor spells out which is which.

Fixed

  • Expanded view and full screen hold steady. Moving between the two, or playing one video after another, used to leave the picture missing until the page was reloaded. Both now give the video the whole pane, every time and in any order.
  • YouTube Music fills the pane properly, with no dark band down its side.
  • Twitch fills it too, with no band along the bottom.
  • The editor stops moving while you tune. The Also applied row used to appear out of nothing the first time a layer went live, pushing everything below it down mid-drag. It now holds its line whether or not there is anything in it.
  • The player says what it refused. A link that went nowhere used to go nowhere silently, which is indistinguishable from a broken page. Refusals, failed loads and the page's own errors are now in the log.
  • Editing several bands at once no longer fails. Every band in a selection sent its own request, and each of those rewrote the whole Equalizer APO configuration. Ten selected bands meant ten rewrites for one turn of a knob, which was slow enough that the edit gave up and reported a timeout — while having worked. A group edit is one write now. Frequency is left out of it on purpose: it is what tells the bands apart, and moving them all by the same number of hertz stacks them on top of each other.
  • The band editor closes when you deselect. It used to fall back to the first band whenever nothing was selected, so clicking away left it open on a band that was highlighted nowhere — and moving a control edited a band you had just let go of.
  • Less memory is used while the graph is open. Some of what the graph allocated was never handed back; less of it is now. This is an improvement rather than a conclusion — the remaining growth is still being tracked down, and closing the graph pane has always released it.
  • The live capture cleans up after itself. Starting it crosses two points where it can be cancelled, and neither was checked, so a capture could survive the request to stop it — along with a timer and a listener that nothing could reach afterwards.
  • Euphoria stops doing work nobody can see. It published a value several times a second that no stylesheet ever read, which made the browser recalculate the whole window's styling for nothing, and it left its glow layers in place after the mode ended.

New

  • A Video tab. A player inside FluidEQ — YouTube, YouTube Music, SoundCloud, Bandcamp, Vimeo, Twitch — so a track can be tuned while it plays, with the spectrum moving underneath it. It is not a browser: it goes to those sites and nowhere else, deliberately.
  • The search box remembers. What you looked for before comes back as you type, closest matches first, with a cross on each to forget it. Anyone checking a crossover plays the same reference track twenty times in a week.
  • Two sizes for the graph, on one menu. Expand view (Ctrl+S) floats it over the workspace while you keep working; Full screen (Ctrl+F) gives it the whole screen and takes the video player fullscreen with it. Escape comes back from either.
  • See through the graph. In full screen, two sliders decide how much of what is behind shows through and how hard it is blurred — all the way to a graph drawn straight over the video with no panel at all.
  • Hide the grid (Ctrl+G), so nothing is left but the wave. Separate from Wave only (Ctrl+W), which hides the EQ curves and keeps the scale.
  • Every shortcut is written down in that menu, beside the thing it does.
  • Loudness. One press, under the euphoria pill, and quiet listening stops sounding thin. The ear loses bass and treble far faster than it loses the midrange as the volume comes down — which is why music at low level sounds hollow rather than merely quiet — so this puts back what your hearing stops picking up. It reads as fuller while the peak level barely moves, and it cannot clip you: the preamp comes down to meet it automatically, the same way it does for every other layer. Not a compressor — Equalizer APO has no dynamics processing at all, and a control claiming otherwise would be lying.
  • Report a problem, in the actions menu. It gathers the logs, strips anything that identifies you, shows you the whole thing, and lets you post it as a GitHub issue or email it privately. Nothing is sent until you press one of those, and you can read and edit every word first. Account names, file paths, email addresses and network share names are removed automatically.
  • Fix audio problems, also in the actions menu. When the sound stops there are four things worth trying, they escalate, and until now they sat in a flat list with nothing to say which to press first or what each one costs you. This puts them in order — restart the audio service, re-select your devices, reinstall Equalizer APO, then remove and re-add the device across two restarts — with what each one fixes and what it will cost, and a button on every step that can be automated. Stop at the first one that works.

Fixed

  • Signing in to YouTube closed the app. Not the tab — the whole thing. The guard that turns back a link leading off the allowed sites was cancelling the navigation from inside the notification that it had started, and taking Chromium down with it.
  • The graph goes on watching after the audio device changes. Reinstalling Equalizer APO, restarting the audio service or switching the default device leaves the capture running against an endpoint Windows has invalidated — live, silent, and never recovering. The trace simply stopped moving. It now notices and re-acquires.
  • The divider between the panes keeps its proportion. It remembered a number of pixels, so a split set on a large monitor was most of a laptop screen — and the correction that kept the panes on screen overwrote the setting, losing it for good. It remembers the ratio.
  • Reinstall Equalizer APO says something useful when the installer is missing. It showed the words apo-bundle-missing over "please restart the application". It now opens Equalizer APO's own download page, which is the one situation where that is the right answer.
  • The window's own errors reach the log. Nothing in the interface had ever written a line to the file a bug report attaches, so a report about the part of the app you actually look at described everything except it. Crashes in the main process are recorded too, rather than leaving a window that just disappears.
  • "Equalizer APO is not responding" can be clicked. It reopens the notice with the Install and Retry buttons — which, once dismissed, previously could not be reached again at all, because it only reappears when the error text changes and for a missing audio engine it never does.
  • The player comes back where you left it after a restart, instead of on YouTube's home page.
  • The live trace is at full strength in Wave only. It was drawn at half strength to sit under the EQ curves — which are exactly what that mode hides.
  • Ordinary limits stop calling themselves internal errors. Adding a band when you already have the maximum, or removing the last one, answered with "Internal Error: Invalid parameter — please reach out to the developers". Nothing had gone wrong and no developer was needed. They now say what the limit is and what to do instead.

Fixed

  • Euphoria mode could not be switched off. Once you had earned it, the toggle did nothing — the mode was reading "your current streak is at the ceiling" as well as the switch, and a streak does not reset when you stop playing. So it stayed true forever and held the colours on no matter what you pressed. Winning is now a moment rather than a state: it unlocks the mode for good and switches it on there and then, and after that the switch is the only thing that decides.
  • The development shortcut invented a score. In development builds, the button that jumps to euphoria used to write a full 19,350 points and a 36-tap streak straight into the run — so the share card could show a number nobody had played for. It now only turns the mode on. Your score is yours.

New

  • Ctrl+E turns euphoria mode on and off, once you have earned it. Before that it does nothing at all: a shortcut that worked early would give away that there is something to find.

Changed

  • The setup log records nothing that identifies you. It was writing the install directory, which on a normal install contains your Windows account name. It now records only what happened — found, offered, declined, started, failed — so the file is safe to send to anyone without reading it first.

Fixed

  • The installer offered to install Equalizer APO and then did nothing. You said yes, no setup window appeared, and FluidEQ opened over the top of it without an audio engine. Equalizer APO's installer requires administrator permission, and the call FluidEQ used to start it cannot ask for it — it failed instantly, and the message saying so went into a pane that one-click installers do not show. It now asks properly, so Windows raises the permission prompt and APO's setup opens. If permission is declined, FluidEQ says so plainly instead of carrying on in silence.
  • FluidEQ could not tell whether Equalizer APO was already installed. Installers run as 32-bit programs and Windows quietly redirects their registry reads away from where 64-bit software records itself, so the check always came back empty. Setup would therefore have re-run APO's installer over a perfectly good installation, and uninstalling FluidEQ could never offer to remove APO, because it could not find it either.
  • Uninstalling could not remove Equalizer APO even when you asked it to, for the same permission reason as the install.
  • The Install Equalizer APO button inside the app was broken the same way. It reported success, and no installer ever opened. Both places now ask Windows for permission properly.

Changed

  • Setup writes a log. Every decision it makes about Equalizer APO — found or not found, offered, declined, started, failed — goes to %APPDATA%FluidEQlogsinstall.log, next to the app's own logs. None of this was visible before, which is why the problem above survived a release.
  • A build missing its copy of Equalizer APO now says so during setup, rather than skipping the step without explanation.

New

  • Equalizer APO comes with FluidEQ. It is inside the installer — nothing is downloaded and there is no second website to visit. Setup offers to run it, and APO's own installer opens so you can choose which audio devices to equalise and get its restart prompt from the tool that owns it. Already have Equalizer APO? It is left completely alone, whatever version it is. This was the worst moment in getting started and it is gone.
  • Uninstalling asks whether to remove Equalizer APO too, and defaults to No. It is a system-wide audio component and other applications may be relying on it, so removing it is never the quiet default. Updating never asks.
  • The Install APO button inside the app runs the bundled copy rather than opening a download page.

Changed

  • The equaliser is properly inert without Equalizer APO. The window still opens, the meter still runs and the graph still draws — the app is worth looking at either way — but every control that writes to the config greys out rather than looking live over an engine that is not installed.
  • The share card opens in its own window. It used to replace the game inside the support panel, which meant a picture the size of the panel trying to fit inside that panel; on a short screen the buttons that save and post it ended up below the fold.
  • The support panel fits a short screen. It gives things up in order of how little they are missed rather than shrinking everything at once — first the spacing tightens, then the aside about how the app was built goes, and last the QR code, since anyone on a window that short is at a desk and the link beside it does the same job in one click.

New

  • The live spectrum can be drawn thirty-six different ways, in two palettes. Bars, a ridge, a staircase, floating peak caps, a contour map that shows how wide a peak is rather than only how tall, a slope field that draws which way the spectrum is turning, stalactites hung from the ceiling, a heart monitor that beats once per band, a car riding the loudest frequency along a road made of the trace, and a rank of little sprites hanging at their levels. Pick one from the searchable list on the legend, or click the graph to walk through them while you listen — Space does the same, and Ctrl with either steps back.
  • Every form moves in its own way. A bar snaps up and hangs, the way a level meter does, so a kick lands as a step. A ridge is a landscape and moves like one, because a hill that twitches reads as noise. The pulse is nearly instant, because a heartbeat that arrives late is not a heartbeat. Choosing a drawing changes how the music feels, not just what it looks like.
  • Wave only hides the EQ response, the voicing and driver layers and the band handles, leaving the live trace alone on the grid.
  • Full screen clears the bands out of the way and presents the graph in the middle of the workspace. Esc brings them back.
  • Ten ways to draw the titlebar meter, cycled by clicking it.
  • Smart EQ is its own layer. It used to rewrite your bands, so applying it overwrote the tuning you had done by hand and clearing it took your work with it. It now sits on top of everything else with its own chip, and survives clearing an AutoEQ reference. It also no longer inverts your own bands: the target it corrects towards includes them, where before it measured as if they were not there.
  • An applied reference arrives band by band, in frequency order, instead of every slider jumping at once. You can see what it did rather than just what it left behind.
  • Zoom the whole interface with Ctrl and +, - or 0. Keyboard only, deliberately — the mouse wheel belongs to the sliders.
  • There is a game hidden in the support panel, for anyone with the contribution badge. Tap the creature or press space on the beat of whatever you are playing. It reads the real percussion out of your own audio, so it is your music you are playing along to.
  • Euphoria mode. Thirty-six consecutive perfect taps and the entire application goes rainbow with the music: the bands, the response curve, the meter, the menus and the release notes. One mistake and it all goes back to being an equaliser.
  • Share your run. A card is drawn with the creature, your score and the mode's own look, ready to post — copy it straight to the clipboard and paste it into whatever you are writing.

Changed

  • Both waveforms move at the display's rate, not the analyser's. Sound is measured about twenty-two times a second, so drawing only when a measurement lands meant two frames in three showed the same picture as the one before — which the eye reads as stepping. They are drawn between measurements now, gliding toward each new one: at the full rate of your screen during euphoria, and capped at thirty otherwise, because a meter beside an equaliser does not need sixty and the frames are not free.
  • The graph keeps time with the music instead of catching up with it. The analysis window was twice as long as it needed to be and the smoothing on top of it added more, so the picture arrived about ninety milliseconds after the sound. It is closer to fifty now, and it is quick to rise and slow to fall rather than sluggish in both directions, which is what makes it pump rather than sway.
  • The graph reaches further across. The axis started at 10Hz but ended at 25kHz, which on a logarithmic scale left three times as much empty space before 20Hz as after 20kHz. Both marks now sit the same distance from their edge, and the drawing gets the space back.
  • The grid and the supporting curves have stepped back so the response you are editing is obviously the subject rather than one line among five.
  • The installer opens FluidEQ when it finishes. Updates already did; a fresh install left you looking at a closed wizard.
  • FluidEQ is called FluidEQ in Task Manager, in development as well as in the installed app.
  • The language picker is one of the app's own dropdowns, not the operating system's, and it opens downwards at a sensible size.
  • The titlebar sits almost flush with the page, so the window reads as one surface rather than a bar stuck on top of one.
  • The delete-band button becomes an icon when the row runs out of room, and gets its label back when there is space for it. It is a bin now, not a cross: a cross reads as "close", which is the wrong promise.
  • The score measures accuracy, not endurance. Only a perfect adds; anything else pays a little and gives back a share of the total. Playing longer cannot move anyone past their own precision.

Fixed

  • FluidEQ was capturing your screen the entire time it was open, and the memory that took never came back — several gigabytes and still climbing after a few minutes. Windows only offers system audio through the same call that offers screen sharing, so a video track arrives whether or not anything wants one. It was being switched off in a way that mutes what a track delivers while leaving the capture behind it running, so full-resolution frames were produced continuously and nothing ever read them. The video is now released outright the moment the audio arrives.
  • Euphoria mode ate memory without limit — a few hundred megabytes with it off, several gigabytes and still climbing with it on. The travelling colour was animated on the window itself and inherited by everything inside it, so the browser had to recompute every element on the page sixty times a second for as long as the mode was running; the interface went sluggish, the graph stuttered, and the waste piled up faster than it could be cleared. The colour now travels only on the handful of things that actually show it. Two of those things also carried an animated glow over a picture that redraws twenty-two times a second, which meant re-rendering everything beneath it on every frame — that glow sits outside the moving part now. It looks the same.
  • A request that timed out leaked its listener, forever. The bridge that cleans up after an internal request rebuilt the wrong function and so removed nothing at all, silently. Beyond the memory, the abandoned listener stayed first in the queue and would answer somebody else's later request with the wrong result.
  • A band deleted mid-animation stayed in memory until the animation it was never going to finish would have ended. One is nothing; an afternoon of adding, dragging and deleting bands is not, and none of it is visible while it accumulates.
  • Clearing an applied reference clears its bands. The notes previously claimed the bands stayed, which was wrong in both directions — a reference is where your tuning came from, not a label beside it.
  • The pet stops reacting to a paused waveform, and stops swaying in a silent room. The analyser freezes mid-frame when paused, and it was dancing to a reading that had stopped being true.
  • The graph could not draw. 0.6.0 shipped a build that dropped d3's transition module, so every animated part of the response chart threw e.transition is not a function and the window came up broken. The build had been told d3 was free of side effects, which is true of almost all of it and false of exactly the piece that matters: transitions install themselves onto d3's selections when the module is evaluated, and nothing refers to them by name, so the bundler quite reasonably concluded they were unused and removed them. 0.6.0 and 0.6.1 were both withdrawn; everything they contained is in this release.

Faster

  • The renderer is 44% smaller. Nothing was being tree-shaken at all — the TypeScript build emitted require() calls before webpack ever saw the import graph, so every unused corner of every library shipped. d3 alone was dragging in geo, force, contour and delaunay to draw an axis.
  • The creature stopped re-rendering twenty-two times a second. It sits in the titlebar, so that never stopped: not when the window was idle, not when nothing was playing.
  • The device list stops polling while the window is hidden. Each tick enumerates every audio endpoint on the machine.
  • The spectrum is drawn as sixty-four columns rather than three hundred and twenty. At that density the bars touched, so it read as a filled area with a ragged top — a worse picture, built from a path string five times longer, rebuilt on every frame.
  • The spellchecker is gone. It was downloading a dictionary to check the spelling of preset names.

New

  • The installer asks which language to use, preselecting your Windows one. Nine of the app's ten, because NSIS has no Hindi translation — FluidEQ still speaks it, the setup wizard does not, which is better than a half-English wizard pretending otherwise.

Changed

  • The release notes open on top of the support dialog instead of replacing it. Reading them is a detour from deciding whether to contribute, not a departure from it, so closing them puts you back where you were.
  • The pet sways like someone listening. It leans, holds there for a beat and comes back, rather than swinging evenly from side to side — and the lean hinges low, where a neck would be, instead of spinning the whole creature about its middle. Still only for anyone with the contribution badge.

Fixed

  • Escape closed both dialogs at once when the release notes were open over the support dialog.

New

  • FluidEQ updates itself. It checks GitHub for a new version, downloads it in the background and offers to restart, with the progress shown in the app rather than as an easy-to-miss system toast. Being offline is not an error and says nothing.
  • What's new. This file, rendered inside the app — shown once after an update, and in the actions menu whenever you want it.
  • The window reopens where you left it, at the size you left it, maximized if it was. The position is only reused if a display still covers it, so unplugging a second monitor cannot strand it somewhere unreachable.
  • No white flash on launch. The window waits until the interface has actually painted instead of appearing blank and filling in.
  • Measurements are capped at what is believable. Published corrections routinely ask for +16 dB at the edges of the audible band, where the rig is measuring its own coupling error rather than the headphone. References are now limited to 12 dB, and 8 dB below 25 Hz or above 14 kHz. Bands you move yourself are untouched — that is your business.
  • Smart EQ shows its work, applying the correction one band at a time in frequency order rather than making every slider jump at once.
  • Clear the applied reference from the AutoEQ panel, the same way you clear a convolution. The bands go with it — a reference is not a label sitting beside your tuning, it is where that tuning came from, and dropping only the name left a curve on screen that nothing on screen accounted for. Your voicing, driver profile and Smart EQ are separate layers and are untouched.

Changed

  • No more profile name box. Renaming happens on the profile row itself, where the name is, with accept and cancel buttons on the field. Two places to type the same thing meant keeping them in sync, and it made Save ambiguous — you could never tell whether it would create a profile or overwrite one. New profile is the only way to create one now, so Save says plainly that it updates.
  • Smart EQ leaves your band layout alone. An earlier build expanded a coarse layout to 31 bands before measuring. Ten bands genuinely cannot describe a measured response, but how many you use is your decision.
  • The applied reference records the measurement, not just the model. Most models have several and they do not sound alike.

Fixed

  • The config adoption added in 0.4.0 could empty the band editor. A block with a preamp and no filters is what FluidEQ writes for a flat EQ, not "the user cleared their bands", and reading it as truth left the EQ page with no sliders at all. Adoption now also refuses when a voicing or driver layer is live, since those reach APO as ordinary filter lines and would otherwise be pulled into the editor as bands and then applied twice.
  • One output's profile could overwrite another's. Two outputs could point at the same profile file, so saving on the speakers silently overwrote the headphones — easy to hit, because "Untitled profile 1" is exactly the name two outputs both end up with. Saving now takes a free name or one this output already owns.
  • Every output keeps a profile. Deleting the one an output was playing through left it with nothing to save to; that case now resets the sound to neutral and creates a fresh empty profile.
  • Renaming a profile to a longer version of its own name aborted halfway. Renaming "Standard" to "Standard 2" cancelled the moment the text passed through "Standard".
  • Changing the band count during Smart EQ did nothing useful. The check meant to notice it could never fire, so the measurement applied gains to bands that no longer existed. It now measures again against the new layout.
  • Dropdowns in the AutoEQ panel were clipped by the equalizer below them.

New

  • The Equalizer APO config is the source of truth. On startup FluidEQ reads the config on disk instead of trusting its own copy of the state. If anything else changed it — a hand edit, another tool, an APO reinstall, a restore from backup — the file wins. What the config cannot express (which voicing, which driver profile, which headphone reference) still lives in the profile, because those reach APO as ordinary filter lines and reading them back as bands would apply them twice.
  • Import EQ settings and impulse responses. An Equalizer APO ParametricEQ or GraphicEQ file, a FluidEQ profile, or any WAV impulse response. The reader understands the full APO grammar, including the pass and notch bands that carry no Gain token.
  • Ten languages. English, 简体中文, हिन्दी, Español, Français, Português, Русский, 日本語, Deutsch, Italiano. Picked from Windows on first run, changeable from the actions menu.
  • The applied headphone is remembered. Applying an AutoEQ reference used to leave nothing behind saying where the curve came from. The model now travels with the profile, is named on the EQ page, and is re-selected in the picker.
  • Everything shaping your sound is named. An "Also applied" strip on the EQ page lists any convolution, voicing, driver correction or headphone reference that is live, each removable in one click.
  • In-app updates. FluidEQ checks GitHub for a new version, downloads it in the background and offers to restart. Being offline is not an error.
  • The window reopens where you left it, at the size you left it, maximized if it was.

Changed

  • The workspace scrolls instead of squeezing every panel into the window. The response graph keeps a readable height on a short window, and the profile list is no longer three rows tall.
  • The named-profile list shows every profile rather than only the one attached to the current output, with the live one marked. New profile creates one for real, named "Untitled profile 1", 2, and so on.
  • The AutoEQ library is collapsible, and states which model is applied even when folded.
  • Icons on the actions menu and the EQ toolbar; only the live band layout is highlighted rather than all four.
  • One shared motion vocabulary: menus unfold from the control that opened them, tab panels rise into place, notices slide in. All of it respects prefers-reduced-motion.

Fixed

  • isAutoPreAmpOn travels with the profile. Turning auto-normalise off on one output turned it off on every output, and the next auto-save wrote that in.
  • Switching outputs reloads everything. The bands, preamp, voicing, driver correction and convolution all belong to the output they were tuned on, but the app only found out about a switch by accident.
  • The EQ toolbar drew over its own heading — eleven controls and no window width at which they fit beside a title.
  • Saving a band with a High Pass or Band Pass filter and reloading replaced the whole EQ with ten default bands. The preset validator only accepted three of the seven filter types.

New

  • Driver-type compensation. Twelve profiles for what you are listening on: dynamic, planar magnetic, balanced armature, electrostatic, bone conduction, the common diaphragm materials, and driver size — each its own layer with a strength slider, drawn as its own curve on the graph.
  • Voicing. Five curated target curves — music, movies, games, speech, late night — written after your bands so your own tuning is never touched.
  • Convolution. Apply a verified minimum-phase impulse response from the AutoEq catalogue before the parametric EQ.
  • Automatic per-output profiles. Edit any control and the tuning saves itself and attaches to the active Windows endpoint. A manual save keeps a separate copy you can always restore.
  • Smart EQ measures what is actually coming out of your output and flattens what it hears.

Fixed

  • Clear EQ left the previous profile fully audible. Equalizer APO accumulates blocks rather than replacing them, so the device block and the session block stacked instead of the newer one winning.
  • One preamp for the whole chain. Every layer used to add its own, so stacking a voicing on a convolution buried the signal.
  • Loading some Squiglink measurements wrote Fc NaN Hz into the config and took the whole chain out.

FluidEQ began as a fork of AQUA and spent its early versions on the device-profile foundation: discovering Windows render endpoints, mapping the stable endpoint GUID rather than the display name, and generating one Equalizer APO block per assignment.

上手的前五分鐘

FluidEQ 入門