The Code of this patch is available on GitHub: https://github.com/Natenom/mumble/commits/opus_music.
(I do not really know if there is a noticeable difference, but I wanted to have this feature :))
'OPUS_APPLICATION_RESTRICTED_LOWDELAY' – “Only use when lowest-achievable latency is what matters most. Voice-optimized modes cannot be used.”
'OPUS_APPLICATION_VOIP' – “Best for most VoIP/videoconference applications where listening quality and intelligibility matter most.”
'OPUS_APPLICATION_AUDIO' – “Best for broadcast/high-fidelity application where the decoded audio should be as close as possible to the input.”
The Mumble client uses the second one (VOIP) to encode the audio input. This cannot be changed in the User interface. To let the user decide whether to optimize for voice or for music (which improves Audio bots) this patch was created.