When the Capture Workflow was first being implemented @jofemodo made this post about audio with more details: