OBS Studio Guides

How to Set Up Desktop Audio in OBS

Choosing between global capture and per application capture is really a question of what you want to exclude, because the failure everyone remembers is a notification baked into a tutorial.

Aug 09, 2026
6 min read
C
Cubix Team

Desktop audio is the sound your computer makes: the application you are demonstrating, the video you are playing, the notification you did not want. OBS can capture it three different ways, and they are not interchangeable. One grabs everything, one grabs a single application, and one is attached to a video source.

Picking the right one is mostly a question of what you want to exclude, because the failure everyone remembers is not silence. It is a finished tutorial with a Slack notification baked into it.

An audio routing diagram showing a capture source feeding a recorder, representing desktop audio capture setup in OBS.

Option 1: Global Desktop Audio (Everything at Once)

The default, and the fastest to set up.

  1. Settings → Audio.
  2. Under Global Audio Devices, set Desktop Audio from Disabled to your speakers or headphones.
  3. Apply, then OK.

Play something. The Desktop Audio meter in the Audio Mixer should move.

What OBS is doing here is a loopback capture of an entire output device, so it hears exactly what you hear. That includes your email client, your calendar reminder, Discord, and the music you forgot was playing. There is no filtering by application.

Two details that catch people out. OBS captures the device you selected, not "whatever is currently default", so switching from speakers to headphones mid session can leave OBS listening to a device nothing is playing to. And if you use one output for your game and another for voice chat, Desktop Audio 2 exists in the same settings panel for capturing the second device.

Global capture is the right choice when you want everything, which mostly means gaming, or when you are recording a video call and need both sides.

Option 2: Application Audio Capture (One Program Only)

This is the option that makes clean tutorials possible, and it is a Windows feature.

  1. Settings → Audio and set the global Desktop Audio to Disabled, so you do not capture the same sound twice.
  2. In Sources, click + and choose Application Audio Capture (BETA).
  3. Name it something obvious, such as "Chrome audio".
  4. In properties, pick the program from the Window dropdown.

You now have a mixer channel carrying only that program. Notifications from everything else are simply not in the capture, no timing luck required.

Worth knowing:

  • It needs Windows 10 version 2004 or newer, and it is still labelled beta.
  • It follows the process you selected. If the application spawns a separate process for audio, or you restart the app, re-check the source. Ticking Match by executable rather than window title in properties makes it stick better across restarts.
  • Some applications, particularly ones that route audio through a system service, do not appear or produce no sound. Global capture with your notifications silenced is the fallback.

There is also a shortcut worth knowing. On recent OBS versions, a Window Capture or Game Capture source has a Capture Audio (BETA) checkbox in its properties. Tick it and that source's audio comes in with its video, no separate audio source needed. It is the same underlying mechanism with less setup.

Option 3: macOS, in One Paragraph

macOS never exposed a system wide loopback to third party applications, which is why every older Mac guide starts by installing BlackHole or Soundflower.

That is no longer necessary on current systems. On macOS 13 Ventura or later, OBS captures desktop audio natively: the macOS Screen Capture source carries the audio of what it is capturing, and OBS 30 and later also offers a dedicated macOS Audio Capture source that can target the whole system or a single application. No virtual driver, no system extension. Below Ventura you still need a virtual audio device.

The Mac path has its own quirks worth reading in full before you set it up, and they are covered in how to record system audio in OBS on Mac, which is the canonical page for that side. The rest of this page applies equally once the audio is arriving.

Enjoying this read?

Get weekly insights on video editing, AI workflows, and creator growth straight to your inbox.

Join Newsletter

The Setting That Makes the File Silent

Here is the failure that wastes the most time: the meters bounce while you record, and the exported file has no desktop audio.

Sources are assigned to audio tracks, and the recording writes only the tracks you enabled. Get those two out of step and the sound is captured but never written.

  1. Settings → Output, set Output Mode to Advanced, open the Recording tab, and note which Audio Track checkboxes are ticked.
  2. Right click in the Audio Mixer → Advanced Audio Properties.
  3. In your Desktop Audio row, make sure a track that is enabled in step 1 is ticked.

While you are there, this is also where you split things out usefully: Track 1 as the mix, Track 2 microphone only, Track 3 desktop audio only. Enable all three in the recording settings and your file carries separate stems, so application audio that turns out to be too loud can be pulled down in an editor instead of re-recorded.

If the meters are not moving in the first place, that is a different problem with its own diagnosis: start at OBS no audio: desktop audio not working, or OBS desktop audio not working on Windows 11 if you are on Windows 11 specifically.

Balancing It Under Your Voice

Once both are arriving, the mix is the last decision, and the default is almost always wrong. Application audio recorded at the level you enjoy while working sits far too high under narration.

  • Narration is the content. Desktop audio is context. Pull the desktop fader down until it is clearly present but never competing, commonly somewhere around -20 to -30 dB relative to where it started.
  • Check it on laptop speakers and phone speakers, not studio headphones. Headphones flatter a bad balance.
  • If you want to keep the option of fixing it later, use the separate track routing above rather than trying to get it perfect live.

For the wider question of getting both feeds into one clean file, recording screen with both mic and system audio covers it tool by tool.

What This Complexity Buys You

Everything above exists because OBS is a broadcast mixer. Per source capture, per track routing and independent monitoring are exactly what you need when you are running a live show with a game, a guest, music and alerts.

For recording a product demo, it is a lot of configuration standing between you and a video, and every step is a chance to end up with a silent file. Cubix Capture takes the opposite approach: it captures system audio and your microphone together with nothing to route and no tracks to assign, and puts its effort into the picture, framing and zooming toward wherever your cursor is working so the recording is done when you stop. It runs on Windows and macOS and the free Starter plan is enough to try the same recording both ways.

If you are streaming and need Spotify out of your VOD while your game stays in, Application Audio Capture is precisely the right tool, and nothing simpler will do that job.

Keep Reading

Don’t miss the next one.

Join our newsletter for exclusive tips, product updates, and the latest from the Cubix team.

Join Newsletter
C

Cubix Team

OBS Studio Guides

Part of the visionary team at Cubix, redefining the future of video creation through agentic AI and seamless workflows.

System Audio With Nothing to Route

Cubix Capture records app audio and your voice together, balanced, with no track assignment and no virtual audio drivers.

Try Cubix Capture Free