Separating game audio and voice over in a video?

I have mine set to Default for Desktop Audio. All game sounds generally got there unless you have some weird set up or changed a lot of settings. If your game sounds don’t go to the Desktop, I’m not sure how to help.

Your current setting is looking for an audio source from your monitor. Unless you directed a sound to your monitor this setting will not work.

This question isn’t really about the separating so if I need to make a new post I will, but since my last video didn’t have any sound, can I use Shotcut to do a voiceover on it?

Here’s a tutorial for it.

Direct show has been renamed to Audio/Video device.


@sauron I managed to do it! Thank you! At least one thing has worked for me!

This is an older tutorial, but has anything improved in the meantime?
I find it incredibly complicated after this procedure, because it doesn’t allow ‘synchr-dubbing’.
For documentation, like in this example, it is quite useful if you have time.

What is synchr-dubbing? I found this pretty easy, but all I was having to do was narrate what was happening in my video, so there wasn’t much to it.

A GUI is available for Spleeter.


Your topic is very useful to me. Thanks a lot

… And very useful to me too …

I did a test with it. The zip includes 4 mp3 tracks:


I took a short piece (Funkorama). Track 1 is the original.
Track 2 - ran it through Spleeter and took off the bass and drums, leaving the instrumental only.
Track 3 - ran it through Spleeter and isolated the bass only.
Track 4 - ran it through Spleeter and isolated the drums only.
I thought the results were very good.

Didn’t have time to test with a piece featuring vocals.

Thanks @sauron for the link to the Spleeter GUI. :+1:

Spleeter test - to split Funkorama into Bass Drums (716.9 KB)

Works pretty well with vocals. (604.5 KB)



Good choice of band. :+1:

I had never heard of spleeter, tried it out and it was a bit convoluted to install
and use on the command line.
Then tried the GUI you posted a link too, much easier but very disappointed with
the resulting stems from spleeter.

Don’t get me wrong, the technology is pretty awesome but the results are terrible.
Too many phase and amplitude variations (must be from the algorithm they are using),
making it sound far too “flangy” and pretty much useless.
There are also plenty of drop outs at times due to the algorithm probably getting confused
and just leaving out small sections.

It can’t be used for any serious work in making remixes or even “remastering” the stems.
Unless all stems are used (i.e. not muted or level increased or decreased by more than 3dB), the horrible flaging sounds just pop up.

Only real use is to use the drum track as a place holder to replace with other drum samples and then deleting the spleeter stem.
These will then always be in time, provided of course one does not use Audacity,
it has terrible timing problems even with the tracks sync-locked, they drift.

Sorry to be so negative about spleeter.
Not a bad start, but it still has a very long way to go to make it useful.

BTW also be very careful when using Audacity for measuring loudness, especially
R128 (with the proper plug-in).
There seems to be some bug where the audio from Audacity being passed to and from the
plug-in varies and the result is completely wrong readings.

In short, thanks for alerting me to spleeter, learnt something today, but for any serious work, it and Audacity are a no go.

This topic was automatically closed after 90 days. New replies are no longer allowed.