/docs · CueLab · Practical

OBS audio routing without surprises

OBS audio routing is more powerful than the default UI suggests. Here is the mental model that makes it predictable.

OBS Studio is the streaming workhorse most people use. Its audio routing is also where most live audio incidents originate — not because OBS is broken, but because its default UI hides choices that matter.

The mental model below makes OBS audio predictable. Once you internalise it, “why is the wrong thing going to my stream” stops being a recurring fire drill.

The three buckets

OBS has three audio destinations:

The stream (what your audience hears)
The recording (what gets saved locally)
The monitor (what you hear in your headphones)

Every audio source can be routed to any combination of these three independently. The default UI only shows you streams; the monitor controls live in Audio Mixer → ⚙ → Advanced Audio Properties → Audio Monitoring.

Until you’ve opened that panel, you don’t actually know what your audio is doing.

The monitoring choices

For each source, monitoring is one of:

Monitor Off: the source goes to stream/recording but not to your headphones. This is the right default for your own microphone — you don’t want to hear yourself with latency.
Monitor Only (mute output): you hear it but stream/recording doesn’t. Good for cue tracks, click tracks, or testing.
Monitor and Output: you hear it and stream/recording gets it too. This is what you want for music beds, browser sources, and anything you need to react to live.

The mistake everyone makes once: setting your own mic to “Monitor and Output”. You then hear yourself with 60–200 ms of delay and can’t talk straight. Set mic to Monitor Off unless you have a specific reason not to.

The Tracks system

In Settings → Output → Audio (Advanced mode), OBS gives you 6 audio tracks. The default streams Track 1, but you can record more.

This is the only sane way to do post-production on recorded streams. Set up:

Track 1 = stream mix (what your audience hears)
Track 2 = host mic isolated
Track 3 = guest mic isolated
Track 4 = system audio / music / browser

Then your recording has separate stems per track. Post-production gets vastly easier because you can normalise, EQ, and balance each speaker independently.

You set this up once per profile. It’s the highest leverage OBS configuration choice you can make.

The browser source trap

Browser sources play their own audio. By default, that audio routes through OBS, mixed with everything else, going to your stream — and to your monitor.

The trap: a tab plays a notification. Your audience hears it. You hear it in your headphones. Now you’re reacting on-stream to a Slack ping nobody else expected.

The fix:

Set browser sources to Monitor Off by default.
Mute the browser source in the OBS audio mixer until you specifically need it.
Set “Page Permissions” on the browser source to deny notifications, autoplay where possible.

This is also baked into CueLab Monitor’s health checks: a browser source unmuted in monitor is flagged as a likely feedback risk.

Push-to-talk and ducking

OBS has push-to-talk built in. It’s buried in the Advanced Audio Properties panel.

Use it when:

Multiple hosts on a single physical mic setup.
You want a mute hotkey for sneezes/coughs.
You’re running an automated music bed and want to duck it under speech automatically.

The ducking part is half-baked in OBS — it works, but it’s not as smooth as a dedicated ducker. For serious shows, use a sidechain compressor (Reaper plugins via VST, or a dedicated audio interface with sidechain) and just route the audio through OBS at the end.

The pre-show check

Every show, every time:

Open Advanced Audio Properties.
Confirm monitor settings per source.
Confirm track assignments.
Run a 10-second test recording. Listen to the recording to verify what your audience would hear is what you intend.

Step 4 is the one most people skip. It’s also the one that catches “I think this is going to stream but actually it’s muted.”