@Iheb Jandoubi AFAIK pronunciation assessment feature can only use audio stream or audio file as its input along with speech config and speech recognizer objects. If you need to use audio from your video, you will have to extract the audio file or stream and then pass it to the pronunciation assessment API. I hope this helps!!
If this answers your query, do click Accept Answer
and Yes
for was this answer helpful. And, if you have any further query do let us know.