For high quality audio I would recommend to use a dedicated streaming tool that does not rely on a web browser for encoding. Reason is that most audio processing and encoding built into modern web browsers is tuned for speech, not music.
Instead, I would use the use the new draft internet standard WHIP for this.
Concretely, I would use Janus Gateway as backend for the video room, e.g. with web frontend Jangouts, and then run command-line tool simple-whip-client
to feed the output of a local music player into the video room.
Opus - one of the audio codecs mandatory with WebRTC - is quite flexible, and can optimize for lower-than GSM narrowband speech, or for higher-than-MP3 wideband music: https://en.wikipedia.org/wiki/Opus_(audio_format)#Quality_comparison_and_low-latency_performance
Benefit of using a dedicated encoder, possible thanks to the WHIP protocol - is that it can be told explicitly to encode using music-oriented Opus encoding (not speech-oriented Opus encoding which is the default with most WebRTC implementations).