Does it support transcribing .flv types of files into text?

[drapeol]

Title

[glangford]

whisper uses ffmpeg to process audio. If ffmpeg can decode it, then whisper should be able to transcribe it. The question is what audio codec is being used in the .flv files.

You can see what formats your version of ffmpeg supports using
ffmpeg -formats

See also

FFMPEG supported media types
http://trac.ffmpeg.org/wiki/SupportedMediaTypesInFormats

StackOverflow - Which are the FLV supported audio types?
https://stackoverflow.com/questions/4716203/which-are-the-flv-supported-audio-types

[drapeol]

Thank you for your reply. I don't know the ffmpeg but you have pointed me in a direction to learn it. I will try to learn and use it afterward.

[a2nath]

Does whisper have a built-in noise rejection/background rejection algo? I'm thinking of adding this feature in my automation script by using the ffmpeg separately and its built-in NR algo if whisper does not support it.

[glangford]

@a2nath No there is no noise rejection used currently. Here is the ffmpeg command used to load audio:

whisper/whisper/audio.py

Line 45 in ba3f3cd

cmd = [