Selective Recording in Streamlabs Desktop lets you choose which scene sources appear in the livestream and which appear in the recording. Audio routing is a separate job: you must decide which applications or inputs reach viewers, which are captured locally, and whether you want separate tracks for editing.
Plan those outputs before changing settings, then inspect both the live view and a test recording. This matters if you want clean gameplay without alerts, need to keep Discord out of the broadcast, or want your microphone and game audio on separate tracks.
Map the stream and recording outputs you want
Write down the result you need in three columns: what you hear locally, what viewers hear in the stream, and what the recording contains. A source can be useful in one place and unwanted in another. An alert, for example, might be audible to you and viewers but absent from a clean gameplay recording. That is not the same configuration as excluding Discord from the stream while still hearing it yourself.
A simple plan for a gameplay broadcast might look like this:
| Sound or visual | You hear or see | Stream | Recording |
|---|---|---|---|
| Game picture | Yes | Yes | Yes |
| Stream alert graphic | Yes | Yes | No |
| Game audio | Yes | Yes | Yes |
| Microphone | Yes | Yes | Yes, on its own track if needed |
| Discord call | Yes | No | No |
Treat the table as a target, not a set of toggles. Streamlabs has different controls for visual source destinations, application capture, monitoring and track assignment. No single feature necessarily implements every row. If you have not settled the stream format or scene layout yet, the guide to continuous Tamil gaming walkthroughs may help you think through the content you want to keep on screen.
Also decide whether this is a live-session setup or a recording workflow that you will edit before publishing. For the latter, separate tracks can make editing easier, while a recording-only clean scene can remove stream graphics. These are distinct choices: selecting a clean picture does not automatically create isolated audio tracks.
Separate visual source selection from audio routing
Selective Recording controls the visual scene sources that go to the stream and the recording. In the editor or Studio view, enable the Selective Recording control in the Sources panel. Streamlabs documents source destination icons there; sources initially go to both outputs, and you can change their destinations individually. Depending on the source and the layout you want, a scene element can appear in the stream, recording, both or neither.
For clean gameplay, keep the game capture in both outputs and disable the recording destination for an alert or overlay that should remain on the stream. You can also use a recording-only camera or alternate overlay if it is available as a source in the scene. Check the stream and recording projector views after changing destinations. Those previews help catch an overlay accidentally removed from the live output or left in the recording.
Do not treat these visual icons as audio routing controls. Removing an alert source from the recorded picture does not by itself establish where its sound goes. Likewise, showing gameplay in the recording does not prove that game audio is on the desired track. Configure the audio path separately and confirm it by listening to the output.
Streamlabs documents a limitation: Selective Recording works only with horizontal sources and disables the vertical output scene, so dual output cannot be displayed. If you need vertical and horizontal presentations together, check the current Streamlabs guidance and the options in your installed version before building the workflow around Selective Recording. This is a layout constraint, not a reason to assume that audio controls behave differently.
If your concern is keeping a prerecorded broadcast running rather than managing a desktop capture session, that is a different operational decision. The article on checking whether a prerecorded stream is live without keeping a PC on covers that separate problem.
Choose which audio applications reach the stream
For an isolated application on Windows, Streamlabs documents Application Audio Capture. Add it as a source, open its properties and select the application or window whose audio you want. If the goal is game audio without Discord, select the game and leave Discord outside that capture path. Add your microphone separately if the stream should hear you.
For this documented isolated-application method, Streamlabs instructs you to disable Desktop Audio in Settings > Audio. Otherwise, the desktop-wide source can bring in sounds you meant to exclude, or duplicate audio already captured by Application Audio Capture. After changing it, play the game and speak into the microphone while watching the mixer. Confirm that the game and mic respond, and that an unrelated desktop sound does not appear in the intended capture path.
This method is for the Windows application-capture scenario described by Streamlabs; do not assume the same control is available or behaves identically on every operating system. Check the current Streamlabs Application Audio Capture guide for the supported workflow and current interface. The exact application names and available capture options may vary with your system and build.
Monitoring is another distinct control. It describes whether a source is played back to you, not simply whether it is routed to the stream. If an alert must be heard by viewers as well as by you, Streamlabs says to use Monitor and Output for that source in Advanced Audio Settings. If an alert is silent on stream, check its output and monitoring mode rather than changing visual source destinations at random.
When you need several different mixes—for example, a sound you hear locally but do not want in either stream or recording—virtual audio devices and a mixer may be relevant. Streamlabs’ advanced routing examples illustrate a VoiceMeeter Banana arrangement, but describe example configurations that may need adjustment. That route can offer more destinations, while adding more points where a signal can be duplicated, sent to the wrong output or lost. Start with the simpler application capture when it meets the target.
Keep selected audio out of a recording when needed
Removing a visual source from the recording and excluding a sound from the recording are not interchangeable. If you want viewers to hear an alert but want a clean local recording, you need to configure its audio destination as well as its visual destination. Decide whether the alert should be audible to you, to viewers and in the file; then follow the controls relevant to each output.
For a basic Windows setup, first establish which audio sources are entering Streamlabs. Application Audio Capture can isolate a selected app, while Desktop Audio can bring in computer-wide sound. If Desktop Audio remains active, it can undermine the intended exclusion by capturing sounds outside the selected application. If the alert is generated as a source within Streamlabs, inspect its Advanced Audio Settings and the monitoring device, rather than assuming it follows the same path as game audio.
A more involved mix may use a virtual mixer. Streamlabs’ routing example sends monitored sources through VoiceMeeter and assigns separate destinations for stream and recording. The important idea is to define the target plainly: what should reach the stream bus, what should reach the recording bus, and what should only be heard locally. Map the example to the device names actually shown on your computer; do not copy a device label or routing diagram as if it were a universal preset.
Pay attention to the monitoring device. In Streamlabs’ example, using the default playback device for monitoring can mix monitored sound back into the recording. If an alert or other sound appears in the file unexpectedly, check whether the monitoring route loops back into a captured desktop or recording input. Change one path at a time, then make a short test recording and listen to it before relying on the setup.
A virtual routing setup is more appropriate when a simple application capture cannot express the different destinations you need. It is also more work to maintain: device names can change, audio applications can choose a different output, and monitoring can create a duplicate path. If you do not need separate local, stream and recording mixes, avoid introducing that extra layer.
Configure separate tracks for editing
Use multi-track recording when the purpose is to edit the microphone, game and other audio independently after capture. In Streamlabs Desktop, open Settings > Output, change Output Mode from Simple to Advanced, then open Recording and choose the recording tracks. Streamlabs documents support for up to six separate audio tracks. In the advanced audio mixer, assign each source to the track or tracks you want.
A practical editing layout might put microphone and game audio on separate tracks, with a mixed reference track if your editor workflow benefits from one. Assign only the sounds you intend to keep to each track. Then check the Streaming tab: it selects the track used for the stream, so that track must contain every sound that viewers should hear. A track plan for editing is not automatically a good live mix.
The recording format matters. Streamlabs states that FLV does not support multiple audio tracks. Its multi-track documentation recommends MP4 or MKV, while another Streamlabs routing guide lists TS, MKV or MP4 as formats supporting multi-track recording. Use a multi-track-capable format shown in your installed build, then verify the file in your editing application. Do not rely on an article typo or assume that a container setting worked merely because the mixer displayed track assignments.
If the editor shows only one track, check that Output Mode is Advanced, the recording tracks are selected, sources are assigned in the mixer and the file format supports multiple tracks. This troubleshooting order separates a configuration issue from a file-container limitation. It is also worth opening the actual recording in the target editor before a long session; the mixer is not proof that the exported file contains the tracks you expect.
If your main concern is stream stability or video rather than audio routing, keep those checks separate. For example, the advice on codec settings when a YouTube stream shows green video addresses a different class of failure and should not be used to infer that audio tracks were recorded correctly.
Check system-audio capture on Mac
Mac system audio needs its own path. Streamlabs’ Mac support guidance says macOS does not let Streamlabs Desktop capture other apps’ audio directly, so a virtual routing tool is needed for system audio. Its documented workflow uses BlackHole and Audio MIDI Setup; it is not the same as the Windows Application Audio Capture or VoiceMeeter instructions.
The broad sequence in Streamlabs’ guide is to create a Multi-Output Device in Audio MIDI Setup containing BlackHole and the built-in output, with built-in output above BlackHole so local monitoring continues. Set that Multi-Output Device as the system sound output, then add BlackHole as a Mic/Auxiliary input in Streamlabs Desktop. Follow the current Streamlabs Mac system-audio guide for the precise interface and steps applicable to your macOS and Streamlabs build.
This arrangement makes system audio available to the capture application while preserving local playback. It also means sounds routed through the system output can enter the capture path, so check what notifications and applications will be audible. Streamlabs’ guide says the optional “Play alerts and sound effects through this device” setting should remain on built-in output or speakers if you do not want system notifications entering the stream or recording.
Do not assume every operating system captures system audio identically, or that a Windows routing diagram maps to a Mac device setup. If local playback stops, or the meter does not move, check the Multi-Output Device selection, the BlackHole input in Streamlabs and the system sound output. Then test with a known sound and listen to the resulting stream and file.
Test the live and recorded outputs
Before a real session, use the projectors to inspect the stream and recording outputs after adjusting Selective Recording. Look for the exact scene elements you intended to include or remove. This catches visual mistakes, but it does not establish that audio is routed correctly. Run an audio test as a separate check.
Make a short recording and play it back in the editor or media player you will actually use. Listen for the game, microphone, alert and any chat or desktop sound. If you configured multiple tracks, solo each one in the editor and confirm that its contents match the plan. A meter moving in Streamlabs tells you that a signal is present at that point; it does not prove the final recording has the intended mix or track structure.
For a live check, use a private or otherwise appropriate test broadcast and listen to the stream output from another device, if your channel workflow permits it. Confirm that viewers hear the intended mix, not just the local monitor. If the stream includes all desktop audio, look again for an active Desktop Audio source when using isolated application capture. If an alert is absent, inspect its monitoring and output settings. If the recording contains a sound you excluded, trace the monitoring and recording routes rather than changing picture selection.
Change one setting at a time and repeat the short test. Keep a note of the working arrangement, including which application is selected, which tracks contain which inputs, and what device is used for monitoring. That note is useful after a software update, a microphone change or moving the setup to another computer. Streamlabs’ setup instructions are documentation, not a guarantee that a particular device combination has been tested on your machine.
If your longer-term aim is a continuously running prerecorded channel, you may not need a desktop scene and local recording process for every broadcast. StreamNeo turns an uploaded video into a YouTube live stream, so the specific pain of keeping your own computer running for a file-based broadcast is removed; it does not replace the audio-routing checks for a Streamlabs Desktop production.
Before committing, compare the operating options on the pricing page. When the file and channel are ready, start free — 24-hour trial, no card.
FAQ
Does Selective Recording keep alert audio out of my recording?
Not by itself. Selective Recording chooses which scene sources appear in the stream and recording; audio routing and monitoring need separate attention. Check the alert’s audio path and test the file.
How do I keep Discord out while capturing a game on Windows?
Use Application Audio Capture for the game, and follow Streamlabs’ guidance to disable Desktop Audio for that isolated capture method. Add the microphone separately if needed, then test that Discord is absent from both the live output and recording.
Why do my separate tracks not appear in the editor?
Check Advanced Output mode, selected recording tracks, source assignments and whether the recording format supports multiple tracks. Streamlabs says FLV does not support multi-track recording; verify the resulting file in your editor rather than relying only on the mixer display.
Can Streamlabs Desktop capture Mac system audio directly?
Streamlabs’ Mac guidance says macOS does not allow direct capture of other apps’ audio in Streamlabs Desktop, and documents a virtual routing workflow using BlackHole. Follow its current Mac instructions and verify local playback, stream audio and the recorded file separately.