STREAMING GUIDE / OBS STUDIO

Make music or game audio dip when you speak in OBS

Keep the music at its usual level between sentences. Use the microphone to turn it down while you speak, and let it rise smoothly when you finish.

START HERE

Add a Compressor filter to the music or game source and choose your microphone as its Sidechain/Ducking Source. The threshold sets which sounds trigger the dip. Attack and release control how quickly the music falls and returns.

Choose exactly what should get quieter

Start with a music bed at the level you want during a break in conversation. Put it on a dedicated OBS Media Source, a browser source with its own audio channel, or a separate app capture. Mute that channel for a few seconds in a test recording. If you can still hear the song, find the second capture path. If the guest or game vanishes with it, separate those sources before adding the filter.

Use a mic that already gives you clear speech. Say a few words and note the name of the source whose meter moves. That is the signal the compressor will listen to. You can use the same setup on a game channel when the game is the sound you want to lower.

Put the compressor on the music and select the mic

  1. In the Audio Mixer, open the music source’s menu, choose Filters, and under audio filters select + → Compressor. A name such as “Music under host” makes it easy to find during rehearsal.
  2. Set Sidechain/Ducking Source to the microphone you identified. Leaving it at None makes the compressor respond to the music’s own level. A compressor on the microphone itself would turn down the voice.
  3. As a starting experiment, try a 4:1 ratio, 50 ms attack, 400 ms release, and 0 dB output gain. Loop the same piece of music while adjusting the threshold for your mic.

Pick a threshold above steady room noise and below ordinary speech. For example, if the room reads around −50 dBFS and spoken phrases reach −18 dBFS, try −30 dB. Speak quietly, then pause and type. Listen for the music to move down for the words and come back during the pause. These numbers are starting points for that comparison.

Use the sentence to tune the recovery

Change one control and replay the same phrase
What you hearWhat to change
Only loud words lower the musicLower the threshold a few dB so ordinary words trigger the reduction. If even a loud word makes no difference, check the selected sidechain source and filter bypass.
The music covers the first syllableShorten the attack. If the drop sounds abrupt once speech starts, lengthen it slightly and compare.
Music swells between syllablesLengthen the release so the reduction lasts across short gaps. Shorten it if the music hangs back too long after a full sentence.
The entire music bed disappearsReduce the ratio or raise the threshold. Listen for the beat to remain audible under speech. Keep output gain at 0 dB while comparing.
The keyboard makes the song dipRaise the threshold and recheck your quietest words. If typing and speech reach similar levels, move the mic closer to your mouth and away from the keys, then recalibrate.

Check the live mix and the editing tracks

Record the same passage with the filter enabled and bypassed. Include music alone, a quiet phrase, several normal sentences with short pauses, and a guest speaking while you stay silent. Wear headphones so the call cannot spill back into the mic. With the host mic as the only trigger, the music should dip for the host and recover under the guest. To duck for the guest instead, select the incoming call as the sidechain and repeat the test. If the music stays low during silence, listen to the trigger source for background noise crossing the threshold.

Listen to the combined mix, then solo the music track in your editor. OBS applies the filter to the source before track assignment, so a music-only track from that source also contains the dips. If you need the original bed for editing, keep the original music file and align it in the editor. For a live-only music source, use separate live and recording track assignments. A quiet passage can still contain music.

After dialing in the dip by ear, drop the rehearsal recording into the live stream LUFS loudness checker to verify that speech plus ducked music stay comfortably below True Peak limits.

Sources & version notes

Documentation checked November 18, 2026. Compressor thresholds and release times depend on your microphone gain and room noise; verify with an A/B rehearsal clip.