Bug: Mass transcription of Video Lectures in German and English, some videos are transcribed only at the beginning

[markdimi]

Hello,

I am trying to transcribe a large collection of lecture videos from difference speakers, in English and German. In a few videos, that are from the same speaker (thus same setup, format, etc), but not all the videos of that speaker, whisper only transcribes a few minutes at the beginning of the video. For example, the videos are almost an hour long and the transcription is done for 6-7 minutes at the beginning.

I run whisper multiple times on the same lectures, but the outcome is the same.

My current pipeline is export the audio from the videos using FFmpeg. Then use the *.wav file (also used MP3s) with standard whisper (large) to transcribe the audio.

Any idea would be greatly appreciated.
Thank you