Multiple audio tracks with output visible in UI but not recording to file

Ivelieu

New Member
First time poster and flatpak user, Void Linux, KDE Plasma 6.1.4 on Wayland (using the flatpak because it has the websockets API for global start/stop hotkeys via pschmitt's obs-cli, and the xbps distribution doesn't - https://github.com/void-linux/void-packages/issues/46835). I want to record some separate audio channels which I can edit after recording, and installed https://github.com/dimtpap/obs-pipewire-audio-capture. I'm running it fine with the `--filesystem=xdg-run/pipewire-0` flag and I am able to add the separate audio sources in OBS, and they correctly light up in the audio mixer in the UI when there's input - so far so good. I go click on the cog at the audio mixer pane for advanced audio properties and set each of the application sources along with my microphone device to separate tracks, then go into settings -> Output -> Recording (simple output mode) -> Audio Track and tick each audio track box I selected in the advanced audio properties window earlier. Try to record - video works fine, but the audio tracks are inaudible. If I bring the mkv into olive, I can see the three audio tracks, and only one (I think the track with microphone) has a visible wave form but it's inaudible.

If I switch to the Output Mode Advanced in the Settings -> Output menu, then select Recording -> Recording Settings -> Audio Track, tick what I'm using, give each audio track a dummy name, then try to record, I get the same result. I think this would accomplish the same thing but I thought to try anyway.

So, the obs-pipewire-audio-capture plugin appears to be working correctly in that it's picking up the application audio, I'm not sure if the issue is with the plugin, something related to the audio tracks or the encoding. All the encoding options are defaults but I did try changing them afterward to no success. In the Simple Output Mode the only audio encoder options are AAC and opus, when I switch to opus I get this error preventing me from starting a recording, on my RTX 4070 Ti SUPER:
init_cuda_ctx: CUDA call "cu->cuInit(0)" failed with CUDA_ERROR_UNKNOWN (999): unknown error
I was not able to get rid of this error by changing the audio encoder back to AAC in the simple menu, deleting everything in /home/darcy/.var/app/com.obsproject.Studio/config/obs-studio/, re-running the auto configuration wizard, restarting my machine, and making a new profile, the only way I could get rid of it was reinstalling the flatpak. But I am able to reproduce this error. I couldn't find anything useful about this error from a google search, no idea if the issue is with the ffmpeg backend command.

Also to note, audio is recorded fine using the default Desktop Audio device if I set it to an audio track, and I can record and hear my microphone device (but NOT the application audio capture sources) only if it is on the same track as a Desktop Audio. In other words if I make a dummy Desktop Audio output, mute it, and add it to the same track as my microphone I can have my microphone and all desktop audio on two separate tracks, but this seems rather hacky and still doesn't let me do application specific tracks.

I tried to find similar issues, the only thing I could find mentioning issues with multiple audio tracks was https://obsproject.com/forum/threads/obs-not-recording-multiple-audio-tracks.149465/, but it doesn't seem useful to me.

The log attached is a session where I make a recording with all the default codecs which gives a mkv with the missing audio.
 

Attachments

  • runlog.txt
    16.4 KB · Views: 17
Top