CLI flag for sentence level (srt) output

[dgoryeo]

Hi,

Thanks again for your great work. A quick rudimentary question: what is the CLI flag to be able to get sentence level srt output?

Thanks

[Jeronymous]

There is no particular flag for this.
If "srt" is an output format (by default with --output_format=all, or with --output_format=srt, or with --output_format=...,srt,...), then two srt files will be generated for each input audio:

• one audio.ext.words.srt with the timestamps of the words
• one audio.ext.srt with the timestamps of the segments identified by Whisper (that might consists on a sentence, a set of sentences, or a part of sentences, depending on the length of the sentence).

If you need to have exactly one sentence per line, you might need to do a specific treatment using audio.ext.words.srt.
I guess you would use punctuation marks to identify ends of sentences (although it might be incorrect sometimes, e.g. with abbreviations "Mr. White is ...").

[dgoryeo]

Thanks for the good clarification @Jeronymous. Very helpful as always.