What options should I use for best results (Pro)?

[smith113-p]

I'll be honest, the models confuse me, and the additional options of pro don't make things simpler.

Something like this?
--standard -f json srt -m large-v2 --ff_vocal_extract mb-roformer --realign -v true

[Purfview]

I'll be honest, the models confuse me, and the additional options of pro don't make things simpler.

Something like this? --standard -f json srt -m large-v2 --ff_vocal_extract mb-roformer --realign -v true

Additionally to your command, setting language instead of auto detection makes result better and add --vad_method pyannote_v3.