Which file goes in which slot?

The program slot takes the audio that should get out of the way, typically music. The sidechain slot takes the audio that should trigger the ducking, typically a voice. They are not interchangeable: putting them the wrong way round ducks the voice every time the music plays, which is the opposite of what is usually wanted.

Why do both files need the same sample rate?

Because the duck is computed sample by sample against the trigger's level. If the two files run at different rates there is no honest way to line them up, so the tool refuses the pair and tells you both rates rather than resampling behind your back and producing something subtly wrong.

What does the lookahead prevent?

It prevents the first syllable of a word from poking through unducked. The program is delayed by the lookahead amount while the gain change is computed ahead of time, so the duck reaches full depth before the trigger's transient actually arrives at the output. Without it, beginnings of words clip over the music.

Why is there a safety limiter?

Because the output is a sum of two signals and adding them can exceed full scale even when neither was loud on its own. The limiter sits at minus 1 dBFS on the mix and is on by default. Turning it off is possible if you intend to process the sum further, but then the level is your responsibility.

Can I export just the ducked music without the voice?

No. The output is the summed mix of the ducked program and the trigger, which is the finished ducked piece. If you need the music alone with the ducking applied, the practical route is to keep the two sources separate and apply the level change yourself in an editor.

Sidechain Compressor

You Might Also Need

Audio Compressor

Dynamic range compression for audio

Parallel Compressor

Blend compressed and dry signals

Dynamic Range Compressor

Professional-grade audio compression

Loudness Normalizer

Normalize to LUFS loudness standards

Sidechain Compressor ducks one audio file whenever a second file is playing, and mixes the two together. Feed music as the program and a voice as the trigger, and the music steps out of the way automatically.

What is Sidechain Compressor?

Load two files into the two slots, set how far the program should be ducked and how quickly it should return, then create the mix. The result is a single file containing the ducked program with the trigger signal mixed on top, and the report states the duck depth, threshold and lookahead used.

The controls and slots:

A program slot for the audio that gets ducked, and a sidechain slot for the audio that triggers the ducking.
A threshold from minus 60 to minus 6 dBFS in single decibel steps, defaulting to minus 30, above which the trigger acts.
A duck range from 0 to 24 decibels, defaulting to 9, which caps how far the program is pulled down.
An attack from 1 to 100 milliseconds and a release from 50 to 1000 milliseconds in steps of 10.
A lookahead from 0 to 20 milliseconds, and a safety limiter on the summed output at minus 1 dBFS, enabled by default.

The tool mixes both files down to mono, pads the shorter one with silence so the two line up from their starts, and then applies gain to the program based on the trigger's level. The duck depth is referenced to the threshold: when the trigger rises above it, the program's gain is pulled down by up to the range you set, and the gain is smoothed by the attack and release times so the movement is not abrupt. Lookahead delays the program by a few milliseconds so that the duck is already at full depth when the trigger's transient arrives, which is what prevents the first syllable of a word clipping through unducked. The two files must share a sample rate, and the tool refuses the pair rather than resampling silently if they do not, naming both rates in the error. Because the output is a mix rather than two files, the effective result is a finished ducked piece rather than a gain curve you can apply elsewhere. The safety limiter protects the sum, since adding two signals can exceed full scale.

Where it is the wrong tool:

It requires both files to share a sample rate, so mismatched material must be converted first.
Both inputs are mixed to mono, so anything genuinely stereo in the program loses its image.
The duck is a single gain over the program, so it cannot duck only the midrange and leave the bass alone.
The output is the summed mix, so there is no way to export the ducking gain on its own.
A long release that suits speech will sound sluggish on percussive triggers.
The two clips are aligned from their starts, with no offset control, so a voice that begins later in the program is handled by silence rather than by timing.

Use it for the everyday voice-over-music problem: a podcast bed, a radio spot, or an interview mixed over an intro. Set the threshold a few decibels below the voice's average, choose a duck range that clears enough space without making the music disappear, and leave the safety limiter on unless you are deliberately mixing hot.

How to use Sidechain Compressor

Load the program into the first slot, which is the audio that will be ducked, and the trigger into the second.
Set the threshold below the trigger's average level, and a duck range that clears enough space for it.
Choose attack, release and lookahead times, using the lookahead so the duck is ready before each word.
Keep the safety limiter on, then create the ducking mix and download the summed WAV.

When to use Sidechain Compressor vs related tools

Sidechain Compressor is for the case where one sound has to make room for another automatically, which is the standard voice over music problem. Audio Compressor is the ordinary dynamics tool when there is only one signal to control, and Parallel Compressor adds density rather than creating space. When the two sources simply need to be heard in sequence, Audio Joiner is the tool for placing them end to end, and Loudness Normalizer finishes the result at a delivery level.

Privacy & Security

This tool runs entirely in your browser — no data ever leaves your device. There is no server round-trip, no upload, no logging, and no account required. Your input is processed locally using client-side JavaScript and is never stored, transmitted, or accessible to anyone else. When you close the tab, everything disappears.