In this guide
Add a compressor to the music source, then select your microphone as its Sidechain/Ducking Source. Test quiet speech, pauses and the music's return in a recording before going live.
To set up audio ducking in OBS Studio, add a Compressor filter to the music or desktop audio source and select your microphone as its Sidechain/Ducking Source. When the microphone crosses the threshold, OBS lowers the target audio; when speech stops, it brings that audio back up. The compressor goes on the sound you want quieter, not on the microphone that triggers it.
This is useful for live commentary over music or gameplay. Ducking changes a source's level; it does not remove music or background noise from a recording. The OBS compressor documentation explains the controls and provides example settings.
Before you start: choose what should duck
In OBS, identify an active microphone source and the source carrying the music. Speak and play music separately while watching the Audio Mixer meters. Name sources clearly so you can tell which signal is which.
If music and game audio both arrive through Desktop Audio, a compressor on Desktop Audio will reduce both. To duck only the music, route it to a separate OBS audio source. If voice and music are already combined in the same incoming source, lowering that source also lowers the voice; look for the original separate signals.
On supported Windows versions, OBS can capture individual applications. Its application audio capture guide explains setup and version requirements. Avoid capturing the same app through both an application source and global Desktop Audio: OBS recommends disabling global Desktop Audio when using application captures to avoid echo. Check that all sounds you need are still captured after changing routing.
Set up sidechain ducking in OBS
- Make a short test scene or save a copy of your setup. Add or identify the microphone and the music source, and make sure each meter moves independently.
- Open filters for the target. In the Audio Mixer, open the music source's menu and choose Filters. Add a Compressor filter. If your intended target is Desktop Audio, open its filters instead.
- Select the trigger. In the compressor, set Sidechain/Ducking Source to the microphone source you actually use. Keep the filter enabled. The music is the target; the microphone controls the reduction.
- Set a starting balance. Adjust your microphone and music to sensible levels first, then try the example settings below. Speak at a normal level while music plays, and listen for the change.
- Record and review. Make a short local test containing quiet speech, a louder phrase and several pauses. Play the saved file back: adjust until speech is clear and the music returns smoothly. Check your stream and recording track assignments if their mixes differ.
OBS compressor settings to start with
OBS's documentation gives the following example for sidechain ducking. These are a starting point, not a guaranteed mix: microphone level, music and routing affect the result.
| Control | Example | What to listen for |
|---|---|---|
| Ratio | 32:1 | How strongly the music is reduced after the trigger crosses the threshold. |
| Threshold | −36 dB | Whether normal and quiet speech trigger ducking without room noise causing unwanted changes. |
| Attack | 100 ms | How quickly the music drops at the beginning of a phrase. |
| Release | 600 ms | How smoothly the music returns after the voice stops. |
| Output Gain | 0 dB | OBS's example does not add output gain for sidechain ducking. |
| Sidechain/Ducking Source | Microphone | Select the actual mic source used in your scene. |
If the music jumps up between words, try a longer release. If the start of a sentence is masked, try a faster attack. If quiet words fail to trigger reduction, adjust the threshold while listening for false triggers from room noise. Change one control at a time and compare the same passage. A gentler ratio or less aggressive threshold may sound more natural for your mix.
Separate music, game audio and recordings
Separate sources let you control what gets ducked; separate recording tracks can also give you more flexibility later. They are related but distinct choices: a separate recording track alone does not split two sounds that entered OBS as one source. Follow the OBS separate recording tracks guide to check assignments and inspect the actual recorded tracks in an editor.
After changing capture sources, play each app in turn and verify its meter, stream mix and recording. Check for duplicates and missing audio. Source names and platform support vary with the OBS version and operating system. See OBS audio monitoring if you need to hear a source in your headphones while testing.
Troubleshoot OBS ducking
| Symptom | What to check |
|---|---|
| Music does not get quieter | Confirm the compressor is enabled on the music target, the correct microphone is selected as sidechain, and speech crosses the threshold. |
| The game gets quieter too | Check whether game and music share Desktop Audio. Separate their capture sources if only the music should duck. |
| Music pumps between words | Try a longer release or less aggressive settings, and review an entire sentence. |
| Speech sounds quieter | Check that the compressor is on the intended target and that the voice is not part of the same captured source. |
| Audio sounds doubled | Look for duplicate Desktop Audio and application captures, or duplicate monitoring paths. Listen to the recorded file as well as your headphones. |
| Volume changes even without the filter | Investigate other filters, app settings and operating-system communications volume behavior separately from the OBS sidechain. |
Can I fix a recording after the stream?
If you saved independent voice and music tracks, adjust their balance in an editor; the audio ducking guide compares manual and automatic methods. With only one mixed track, a volume change affects the combined sound. An original isolated microphone recording is preferable when available.
For a saved file with unwanted background sound, you can test the Saturalabs voice isolator separately from OBS. It processes an uploaded recording, not the live stream; separation can change speech or leave unwanted sound. AI WAV downloads require an account and sufficient credits. Compare a short sample and retain the original.
Common questions
Should the compressor go on the microphone?
For music ducking, put this compressor on the music or other target source and choose the microphone as the sidechain. A separate compressor on the microphone serves a different purpose.
Does ducking remove background music?
No. It temporarily lowers the selected source in response to the trigger. Music remains in the output.
Is Windows audio ducking the same setting?
Windows communications volume changes are operating-system behavior; the OBS sidechain compressor is a filter on an OBS source. If volume changes unexpectedly, check each independently.