How to Set Up Multiple Audio Tracks in OBS

A single mixed audio track is a decision you make at record time and cannot undo. If the application you were demonstrating turned out to be twice as loud as your narration, or a notification landed in the middle of a sentence, there is no version of that file where the two are separable. They were summed before they were written.

Multiple audio tracks push that decision to the other side of the recording. Your voice, your system audio and anything else land in the same file as independent streams, and every balance choice stays open until you are in an editor with the whole take in front of you.

Three separate settings screens have to agree before that works, and one property of media players causes the most common follow-up question. This page covers all four.

What a Multi Track Recording Actually Is

OBS mixes internally to six numbered buses. Every audio source can be sent to any combination of them, and the recording writes whichever buses you told it to write, each as a separate encoded stream inside one file.

This is not the same as a stereo recording with the voice on the left. Each track is a full stereo pair with its own encoder and its own bitrate. A four track recording contains four independent pieces of audio and one piece of video.

Two consequences fall out of that, and both matter later:

• Nothing is separated unless you separate it. The default routing sends every source to every track, so an untouched six-track recording is six identical copies of the same mix.
• Not every container can hold them, and not every player will show them.

Step 1: Write the Tracks

Go to Settings → Output, put Output Mode on Advanced, then open Recording. You will find six numbered Audio Track boxes.

Tick the ones you want in the file. Everything else is still mixed and still metered, and then discarded as OBS writes, which is the mechanism behind most reports of a silent recording.

Each ticked track has its own bitrate. 192 kbps per track is a sensible recording value; there is no reason to spend more on a track that only carries speech.

Step 2: Pick a Container That Can Hold Them

This step is skipped constantly and it is where multi track setups quietly fail.

Use Hybrid MP4 if your OBS version offers it. It writes progressively, so a crash leaves you with everything recorded up to that moment, and it produces a normal MP4 that carries all of your tracks.

On an older build, record MKV and change the container afterwards. That is a rewrite of the wrapper, not a re-encode, so every track survives untouched. File → Remux Recordings does it without leaving OBS; the remuxing walkthrough has the details and the failure cases.

Never record plain MP4 for anything long. Its index is written when the file is closed, so a crash at minute fifty leaves a file with no index at all, and both audio and video are usually gone with it.

Step 3: Assign Sources in the Matrix

In the Audio Mixer, right-click and pick Advanced Audio Properties. One row per source, and six track boxes at the end of every row.

Every box is ticked by default. That is why an unconfigured multi track recording produces six copies of the same mix and looks broken. Separation only exists where you untick.

Here is a layout that works for almost everyone:

Track 1 stays the full mix, and tracks 2 to 4 are the isolated stems.

Why keep track 1 as the mix

Two reasons, and the second is the one people discover the hard way.

Streaming reads a single track, chosen in Settings → Output → Streaming, and that is conventionally track 1. Keeping everything on it means a live audience hears the complete mix.

More importantly, track 1 is what a casual viewer will hear. Most players, most upload pipelines and most preview windows read the first audio track and ignore the rest. If track 1 is empty because you separated everything perfectly onto tracks 2, 3 and 4, your file plays silent for everyone who does not know to go looking.

Why the File Seems to Have Only One Track

This is the most common follow-up, and nothing is wrong with the file.

Playing a multi track file usually plays exactly one of its tracks, because that is what players do. Different tools handle it differently:

• VLC plays the first track. Audio → Audio Track switches between them, and this is the fastest way to confirm a recording worked.
• QuickTime on macOS exposes alternate tracks in the movie inspector.
• Windows Media Player and most browser video elements play the first track with no way to change it.
• YouTube and most upload pipelines take the first track and discard the rest.

So a multi track recording is a working file, not a delivery file. The version you publish should be a normal single track export made after the mix is settled.

What Your Editor Does With It

The point of all this is what happens on import.

• DaVinci Resolve imports every track. Clip attributes let you map them to timeline audio tracks, and the free version handles this fine.
• Premiere Pro imports the tracks and exposes them per clip. If you only see one, check the audio channel mapping on the source clip rather than the file.
• Audacity imports through the FFmpeg library, which has to be installed separately.
• Adobe Audition opens all tracks directly and is the most direct route if you only care about the audio.

The workflow this makes possible is the whole reason to bother: pull the application audio down 8 dB under the narration, mute the notification that landed at 4:12, and export. None of that requires a second take.

The Cost

Multi track recording is cheap but not free, and it is worth knowing the numbers before you enable all six.

File size. Each track is encoded independently. Four tracks at 192 kbps is 768 kbps of audio, about 345 MB per hour. Next to a 1080p60 video stream that is small, but on a long session it is real.

CPU. Each track is a separate AAC encode. It is a minor load compared with video encoding, though on a machine already showing encoding overloaded warnings, six audio tracks are not the place to spend headroom.

Attention. Every extra track is another row in the matrix that can be wrong, and a mis-ticked box produces a file that is silent in a way you will not notice until you open it. Three tracks covers almost every real case. Six is a streaming setup with guests, music beds and separate alert routing.

Before any long session, run the ten second sanity check: record briefly, open the result in VLC, and step through every track.

When One Track Is the Right Answer

Separate stems exist because a live broadcast mixer has to serve several destinations at once, and because a recording of several people cannot be balanced in advance. Both are real problems and OBS solves them properly.

Recording yourself walking through a product is not either of those problems. There, multi track routing is insurance against a mix you were only going to make one decision about anyway.

Cubix Capture starts from the view that one person recording one screen should never see an audio matrix. Voice and system sound arrive together and already balanced, and the effort goes into the picture instead, moving the frame toward whatever you are clicking. The take is done when you stop, not when the mix is settled. Windows and macOS, free tier available.

If you are recording a podcast with three guests, or a stream where Spotify has to stay out of the VOD while your game stays in, the track matrix above is exactly the right tool and nothing simpler replaces it.

Keep Reading

• From capture to file, including the test that catches a silent export early.
• What each capture source actually hears, and what it quietly includes.
• The container question in full, beyond the summary table above.
• Changing the wrapper without re-encoding once the recording exists.
• Getting the voice right in the first place before splitting anything out.