I have also worked with pyonnate but it doesnt detect properly who is is speaking , it gets confused between different speakers. so i go for fixing the variable num_speaker but there is a usecase that we dont know total number of speaker in audio gonna be . Can you please guide on this ?