-
If I use this stable-ts modification, the word containing diacritics will be split. If I use the original whisper the diacritics are not a problem. Maybe a unicode problem. Any ideas for a solution? [stable-ts] - Word by word example: 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 [stable-ts] - JSON example: [original-whisper] - Word by word example: `1 2 3 4 5 6 7 8 9 |
Beta Was this translation helpful? Give feedback.
Replies: 2 comments
-
Since not all languages uses space to separate words it was not made default (except English). But you can specify this behavior with from stable_whisper import results_to_word_srt
results_to_word_srt(results, 'audio.srt', combine_compound=True) # strip=True to remove the space before the first word
|
Beta Was this translation helpful? Give feedback.
-
English works well. Other languages have a problem. For other languages, if combine_compound=False, the word will be split into consonants and vowels. If combine_compound=True instead of outputting one word and its timestamp is 2 or 3 words. Good work. Keep going |
Beta Was this translation helpful? Give feedback.
Since not all languages uses space to separate words it was not made default (except English). But you can specify this behavior with
combine_compound=True
forresults_to_word_srt
orgroup_word_timestamps
.