Implementation with Large-v3 but with Batching

[souvikqb]

I saw a large-v3 implementation with faster_whisper (#547) but it's quite slow.

Large-v3 is very fast with batching as shown here --- https://huggingface.co/openai/whisper-large-v3

Batching speeds up the transcription process by a lot. The only reason I wish to use faster_whisper is cause it provides things like srt, verbose, word level transcription

[phineas-pta]

srt, verbose, word level transcription

u can write code to do those things with transformers interface

[souvikqb]

srt, verbose, word level transcription

u can write code to do those things with transformers interface

Do you have any resources that I can use?

[phineas-pta]

word timestamp is an option when u call the pipeline

fort srt just retrieve timestamp and format properly

[souvikqb]

word timestamp is an option when u call the pipeline

fort srt just retrieve timestamp and format properly

No I meant do you have a dedicated function that I can use to convert it? For srt and verbose

[phineas-pta]

see my comment in openai/whisper#654

[souvikqb]

see my comment in openai/whisper#654

Hey thanks! That sorts out my srt requirement. Now just looking for verbose and word

[phineas-pta]

as i said, word timestamp is an option when call the pipeline

[souvikqb]

as i said, word timestamp is an option when call the pipeline

I actually got an error on running this --

result = pipe(sample, return_timestamps="word")
print(result["chunks"])

[phineas-pta]

ask in transformers github repo