You could potentially set up a GSPipeline to combine the video from some source (like a webcam) with the audio being played at the moment into a single output file, as it is explained here.

But with this method you cannot use the output of the sketch as source. I think this requires an enhanced GSMovieMaker class with sound support. I'll add this feature request to the tracker (but I won't be able to implement it right now).