Arabic diarization

[OsamaRyees]

salam,
I am looking for the best SOTA diarization for arabic and seen your projects. I am using whisper large with whisperX as a start but i fail to get it right,
I also tried to run the colab with the same setup just replacing the audio file but it detect them all as 1 speaker. it detect it is in arabic though
did i do something wrong or should i tweak something on the collab at certin params to get it right
if arabic is not well supported, could you point me out for the best diraization model to use with fosha/dialects

your help is greatly appreachated
regards

[MahmoudAshraf97]

Alaikum Assalam,
Diarization is mostly language agnostic, there's a known issue with very long audios that it attributes all the text to a single speaker، can you try cutting the audio into 30m segments?