When I use whisperx transcribation I have ideal transcribation but diarization very bad. When I use diarize.py I got excellent diarization but very bad transcribe. But in both case used whisperx transcribe model. Maybe problem with aligment which realized throw deepmultilingualpunctuation?
It is possible to use whisperx transcribe and HF allign and then just diarize with Nemo without deepmultilingualpunctuation?
When I use whisperx transcribation I have ideal transcribation but diarization very bad. When I use diarize.py I got excellent diarization but very bad transcribe. But in both case used whisperx transcribe model. Maybe problem with aligment which realized throw deepmultilingualpunctuation?
It is possible to use whisperx transcribe and HF allign and then just diarize with Nemo without deepmultilingualpunctuation?