How to Set Up Automatic Music Ducking in OBS

How automatic music ducking works in OBS

To make music quieter during speech, OBS uses a compressor with an external control signal. The compressor is applied to the music source, and the microphone is specified as the sidechain, or ducking, source. When the voice level exceeds the set threshold, OBS lowers the music volume. After a pause, the music gradually returns to its previous level.

This method is called ducking. The music does not turn off completely, so the stream keeps its background, but the streamer's words become more noticeable. If you need to remove the music instantly every time the microphone turns on, it is more convenient to assign a hotkey to turn off the source or change the volume manually.

For this to work properly, it is important to separate the voice and music into different sources. If music, game, and system notifications all end up in one Desktop Audio channel, the compressor will duck that entire channel rather than just the music.

Prepare separate audio sources

First, find the microphone and music in the OBS audio mixer. Music can come from a media source, browser source, VLC, application capture, or another separate source. The microphone usually appears as Mic/Aux or as an input audio capture source.

OBS 28 and newer include a built-in Application Audio Capture source. On systems where it is available, it simplifies audio separation: you can select a specific application, such as a music player or browser, and route its audio to a separate OBS source. Then you won't have to mix the music with the entire desktop volume.

Add this source using the "+" button in the Sources panel, select the application, and check that its audio appears in the audio mixer. If the selected program is still also going into Desktop Audio, disable the duplicate route. Otherwise the same music will play twice, and the compressor will only duck one copy.

On other operating systems and in some OBS builds, the set of available sources may differ. In that case, use a media source, browser source, VLC, or another method that lets you get a separate music channel. If the music and game are already combined in Desktop Audio, you won't be able to duck only the music with the standard compressor.

Put on headphones before setting up. Check that the microphone is not returning to them through an extra monitoring route, otherwise you'll get echo. In the advanced audio properties, make sure the music is not added to the scene twice.

Add a compressor to the music source

Open the OBS audio mixer and click the gear icon next to the music source. Select "Filters" and add the "Compressor" audio filter. The filter must be applied to the music, not the microphone. Otherwise the voice will be compressed, but the music source volume won't change.

In the compressor settings, find the "Sidechain/Ducking Source" field and select the microphone. The parameter name may differ slightly depending on the OBS interface language. The list should show the source where your voice appears, not the music source and not the overall Desktop Audio.

As a starting setup, you can use these values:

• threshold — approximately -18 to -12 dB;
• ratio — around 4:1;
• attack — approximately 10–30 milliseconds;
• release — approximately 250–500 milliseconds;
• output gain — 0 dB.

These are not universal parameters. They depend on the microphone level, music volume, and the nature of the speech. The main thing at this stage is to select the microphone in the sidechain field and make sure the filter is applied to the correct source.

Set the threshold and ducking depth

The threshold determines at what level of the control signal the music volume will start to decrease. Start speaking in a normal voice and watch the microphone indicator in OBS. If the music only reacts to loud phrases, lower the threshold. If it ducks from breathing, keyboard clicks, or background noise, raise the threshold.

The ratio determines how much the compressor will reduce the music after the threshold is exceeded. At a low value, the voice is only slightly highlighted against the background. At a higher value, the music goes down more noticeably, but too aggressive processing can create the feeling that the volume is constantly "breathing."

Attack controls how quickly ducking starts. A small value frees up space for the first words faster. With too long an attack, the beginning of a phrase may come through over music that is still loud. Release sets how quickly the music returns. Too short a value creates the effect of sharp jumps, while too long a value leaves the background quiet even after the phrase ends.

Don't try to compensate for the result by increasing the output gain. It can bring the music back to its previous volume and reduce ducking to a minimum. First set the threshold, ratio, attack, and release, then evaluate the overall balance by ear.

Account for noise and microphone filters

The compressor reacts to the signal it receives from the sidechain source. If the microphone constantly picks up a fan, computer hum, or keyboard noise, the music may stay quiet even during silence. In that case, first reduce the noise in the microphone source itself, and only then adjust the compressor on the music.

For the microphone, you can use noise suppression or an expander. If you apply multiple filters, the processing that removes background noise is usually placed before the microphone compressor. This way the control signal will react less to extraneous sounds. Don't set overly aggressive values: the voice may become unnatural or cut out on quiet words.

Check that the microphone is not disabled, not muted in the mixer, and is actually sending a signal to the selected source. If the required microphone is not in the sidechain list, check that it has been added to the scene and check the source name. Sometimes people select one microphone but speak into another one connected through a global audio input.

Also make sure the voice isn't reaching OBS through multiple routes. Two identical microphone sources can increase the control signal and cause the music to duck more than necessary.

Check the result on a short recording

Don't evaluate the setup only by the indicators in OBS and don't wait for your first stream. Make a short local recording with music and say phrases several times in a normal voice, loudly and quietly. Leave pauses, laugh, tap on the table, and talk a little over music peaks. This will show how the compressor behaves in different situations.

Listen to the recording in full after stopping. In real time, attention switches between speech, gameplay, and settings, so sharp volume jumps can go unnoticed. On a recording, it's easier to determine whether the first words are audible, whether the music stays too quiet after a pause, and whether it returns with a jerk.

If the music isn't ducking, check the filter on the music source, the presence of the microphone in the sidechain field, and the microphone level itself. If the threshold is too high, the voice doesn't reach the trigger point. If the music stays quiet all the time, the cause is most often background noise, too low a threshold, or excessive microphone signal volume.

If the first words are hard to hear, reduce the attack time. If the music cuts back in sharply between words, increase the release time. A less aggressive ratio and a smoother volume return also help with noticeable "pumping."

Listen to the recording not only in headphones but also on another device at normal volume. In headphones, it's easy to mistake deep ducking for a comfortable balance, whereas on a phone or speakers the voice and music may be perceived differently. Do the final check at the volume level at which streams are usually watched.

Common ducking setup mistakes

The most common mistake is adding the compressor to the microphone instead of the music. The compressor will indeed work, but it will change the dynamics of the voice. For ducking, you need to open the filters of the music source and use the microphone only as a control signal.

Another mistake is selecting Desktop Audio as the music source when the game, browser, and system sounds are already going into that channel. As a result, all these signals will drop at the same time during speech. Separate the sources as much as your OBS configuration allows. In OBS 28+, it's convenient to check the "Application Audio Capture" source for this.

Don't add multiple identical compressors unless necessary. They can cause too deep ducking and unpredictable recovery. Also check that the music isn't being output in parallel through another source: the second copy will stay loud even if the first is processed correctly.

If you use application audio capture, don't leave the same player in the general system capture unless necessary. Duplication can cause echo, delay, or a feeling of excessive volume. After changing routes, make another short recording.

For talk streams, automatic ducking is usually more convenient than moving the fader manually. But if the music plays rarely or needs to turn off completely, use hotkeys and scenes. The compressor is better suited for constant background audio that should remain audible during speech.

Quick step-by-step order

1. Separate the microphone and music into different OBS sources.
2. In OBS 28+, if the feature is available, add "Application Audio Capture" and select the program with the music.
3. Remove the duplicate audio of that program from Desktop Audio.
4. Open the filters of the music source and add "Compressor."
5. Select the microphone in the "Sidechain/Ducking Source" field.
6. Start with moderate threshold, ratio, attack, and release values.
7. Make a short recording with speech, pauses, and background noise.
8. Based on the result, adjust the threshold, depth, and speed of the music's return.

After checking, the music will automatically recede into the background when you speak, but will remain part of the stream's sound atmosphere.