Audio Equalizer Online – 10-Band EQ That Hits Its Numbers

Why use an equalizer

A 10-band graphic equalizer whose sliders are accurate. Set a band to +6 dB and the finished file measures +6 dB at that frequency — which is not what happens on a typical online EQ, and the response curve above the sliders shows you why. Everything runs on your device.

An equalizer shapes tone by lifting or cutting specific frequency ranges. It is how a thin recording gets body, how a boomy one loses mud, and how a voice buried under a room gets its consonants back. What makes it hard to use well is that you normally cannot see what you are doing — so this tool draws the response your file will actually receive, over the spectrum your file actually has.

Music: add low-end weight, open up the top, or start from a genre preset and adjust.
Voice and podcasts: cut rumble with the high-pass, lift presence around 2 to 4 kHz, take out boxiness.
Video and content: match the tone of clips recorded on different days or different devices.
Playback correction: compensate for headphones or speakers that are heavy or thin in a known range.

The curve you draw is the curve you get

A graphic equalizer is a chain of filters, one per band, and each one affects its neighbours. Their effects add up, so the response between and at two boosted bands is higher than either slider says. This is true of every browser equalizer we measured: two adjacent bands at +6 dB produce about +7.15 dB, and all ten at +6 dB produce a tilted curve reaching +9.6 dB. Nobody shows the composite response, so nobody notices.

This tool solves the whole chain instead. It works out what each band has to do so that the combined response passes through the value you asked for at every band centre, then applies that. Measured on the finished files across every preset and several deliberate extremes, the error is 0.00 to 0.02 dB where the ordinary approach is 1.15 to 3.55 dB. The solve also folds in the bass and treble shelves, so the band sliders keep their meaning when a shelf is running. You can turn it off from the rail and watch the curve pull away from the slider dots.

The picture above the sliders

Three things are drawn on one set of axes: the response the file will get, the targets you set as dots, and — once you analyse the file — its own spectrum in the background. That last one matters more than it sounds. Boosting 63 Hz on a recording with nothing below 100 Hz raises noise and nothing else, and the spectrum is what tells you before you spend the headroom rather than after.

Boosts that cannot clip

Lifting a band raises the level, and enough lifting pushes the file past full scale, where it distorts. The clip guard measures the equalized signal's true peak, subtracts the ceiling you set, and applies exactly that much attenuation — no more, so the file is not left quieter than it needs to be. The finished file is then re-measured and checked against the ceiling before it is handed over. Across every case tested it lands on the ceiling exactly.

Three ways to work

10-band graphic: one-octave centres from 31.5 Hz to 16 kHz. The everyday tool, free on both plans.
31-band third-octave: 20 Hz to 20 kHz on the grid a hardware graphic EQ uses, for narrow corrections. Pro.
6-band parametric: place each band where you want it, with your own Q and filter type. Pro.

How it works

Filters run in order: high-pass and low-pass first, because they remove material and everything after should shape what survives; then the shelves; then the bands; then the level correction the clip guard worked out. The file is decoded, filtered and re-encoded once, on your device, not our servers.

A 10-band graphic EQ with eighteen presets, free
A live response curve and your file's own spectrum
Sliders whose numbers the finished file actually measures
High-pass, low-pass, bass shelf and treble shelf
Clip-safe output, verified against the file you download
31-band, parametric and Auto-EQ with Pro
MP3, WAV, FLAC, OGG, AAC and M4A, or keep the source format
A batch of 25 files and one ZIP with Pro

Formats and what happens to them

Audio files go in — MP3, WAV, FLAC, OGG, Opus, M4A, AAC, AIFF, WMA and more — and so do video files, whose audio track is taken out and the picture discarded. Output is MP3, WAV, FLAC, OGG, AAC or M4A, or the format you started with. If every band is flat and nothing else is set, the tool does not re-encode at all: the original frames are copied straight through, which is both instant and bit-identical. Artwork is not carried across; tags are.

Tips and best practices

Cut before you boost: removing what is in the way usually sounds better than adding more of what you want.
Analyse first: the spectrum tells you whether the frequency you are reaching for is actually in the file.
Keep moves small: 3 to 6 dB does most real work. ±12 dB is there for correction, not for taste.
Keep the source format: if you do not need a different container, a flat setting will not re-encode at all.
Leave the clip guard on: it costs one extra measurement pass and makes a heavy boost safe.

Frequently asked questions

Why does my other equalizer sound louder than the number on the slider?

Because a graphic equalizer is a stack of overlapping filters, and their skirts add together. Set two neighbouring bands to +6 dB on a typical online EQ and the response between them reaches about +7.15 dB, not +6. This tool solves the whole cascade so the composite response passes through the value you set at every band centre — measured at 0.00 to 0.02 dB against 1.15 to 3.55 dB for the usual approach. You can switch the solve off to hear the difference.

What EQ settings make a podcast voice sound clearer?

Start with the Vocal preset, add a high-pass filter around 80 to 100 Hz to lose room rumble, and lift 2 to 4 kHz slightly for presence. If it sounds boxy, cut a little at 200 to 400 Hz. Analyse the file first — the spectrum drawn behind the curve shows whether the low end you are about to cut is actually there.

How do I boost bass without making it muddy?

Lift 63 to 125 Hz for weight and leave 200 to 315 Hz alone or slightly down, since that is where mud lives. Keep boosts to 3 to 6 dB. The clip guard measures your file after the EQ and pulls the level back by exactly the amount needed, so a bass boost cannot push the file into distortion.

What is the difference between the bands and the bass and treble shelves?

Each band affects a one-octave range around its centre. A shelf lifts or cuts everything below or above its corner in one broad move, which suits overall tonal balance rather than a specific problem. The shelves sit outside the solve, so a shelf still shapes the sound when every band is flat, and the band sliders keep meaning their own contribution on top of it.

Will heavy boosts make the file clip?

Not with the clip guard on, which is the default. The tool measures the equalized signal's true peak, works out exactly how far it exceeds your ceiling, and applies that much attenuation — then re-measures the finished file and refuses to hand it over if it is still above the line. With the guard off nothing is attenuated and the tool warns you what the curve's peak gain is.

Can I fix a recording that sounds like a big echoey room?

An equalizer cannot remove reverb, but it can take the edge off the harshness room reflections add. A small cut between 500 Hz and 2 kHz reduces honk and boxiness. For the echo itself, use the noise reduction tool first, then come back and set the tone.

What does the 31-band mode give me that the 10-band does not?

Third-octave resolution: 31 centres from 20 Hz to 20 kHz instead of 10 from 31.5 Hz to 16 kHz, which is the grid a hardware graphic EQ uses. It is for correcting a narrow problem — a room mode, a resonance — that a one-octave band would smear across everything around it. Parametric mode goes further and lets you place six bands anywhere with your own Q.

Does equalizing change the file format or quality?

Only if you ask it to. Keep the source format and the tool writes the same container. If every band is flat and nothing else is set, it does not re-encode at all — the original frames are copied through, so the result is bit-identical and takes milliseconds. Otherwise it re-encodes once, at the bitrate and sample rate you choose.