Seven workspaces, one signal chain
EQ, AutoEQ, Voicing, Convolution, Media and Karaoke, with Config off to one side. Every layer reaches Equalizer APO as a file of its own, so any of them switches off without being taken apart — and everything active is named on the EQ page, so a bump in the graph is never a mystery.
Every setting belongs to the device you tuned it on. Plug in your headphones and their tuning comes back; switch to speakers and theirs does. Mapped to the stable Windows endpoint ID rather than the display name, so it survives renames and re-plugs.
Each layer is written as its own Equalizer APO file, so switching one off leaves it out of the chain instead of clearing it. Press the switch again and exactly the same thing comes back — after a restart too. Every correction has a strength, because half of a measured fix is a reasonable thing to want.
6,028 headphone models and 8,850 responses ship offline from the official AutoEq results, with Squiglink available online as a second source. The correction is a layer of its own — its own chip, strength, switch and curve — so clearing your bands no longer throws it away with them.
Measures what is actually coming out of your output and flattens what it hears. As a mode it keeps going — measure, move a fraction of the way, measure again — so one track's emphasis and the next one's cancel out and only what every record agrees on survives.
Music, movies, games, speech and late night, plus eight tuned by genre — rock, metal, pop, hip-hop and R&B, electronic, jazz, classical and acoustic. Each has a strength slider, and each is written after your bands so your own tuning is never overwritten.
A measured impulse response applied ahead of everything else — verified minimum-phase responses from the AutoEq catalogue, or any WAV of your own. It has a switch of its own now, so trying the room without it is one press.
Open a song and its lyrics — LRC, eLRC or UltraStar — and FluidEQ builds a stage: timed words, a pitch lane showing whether you are high, in tune or low, a melody guide tone, guitar chords estimated from the backing track, and a review afterwards naming the parts worth practising. Your microphone is analysed on your own machine and never recorded.
Mirror what you are hearing to a second device, with a level for each and nothing to install. It arrives about a fifth of a second late, so it suits music in another room rather than video, and it runs only while FluidEQ is open. Said plainly here because it is said plainly in the app.
A Config tab shows what Equalizer APO has actually got, not what FluidEQ intends — and the two differ exactly when it matters, after a hand edit or another tool. Each output also gets a file FluidEQ creates empty and never touches again, applied last, for APO commands with no interface here.
Every layer contributes to a single preamp value derived from the real combined response, so stacking a voicing on a convolution cannot clip you. It gives volume back as well as taking it away — though it will not invent any, because away from a narrow cut the chain never moved.
Both channels separately, in true dBFS — nothing on screen answered "how loud is this, really" before. The title-bar wave draws the same measurements ten different ways, and the graph legend is a set of switches, one per curve.
Every label, hint, error and tooltip is translated — with a test that fails the build when one falls behind English. The installer offers a language too, preselecting your Windows one.
Local and account-free. No cloud, no telemetry, no proprietary driver, no virtual audio device.
The EQ workspace
AutoEQ
Voicing
Convolution
Media
Karaoke
Karaoke, singing
The visualiser, editable