Open akashmamun610 opened 1 year ago
Did you use --no_align
?
Thanks for your message.. My input was
!whisperx "/content/drive/MyDrive/Dvd" --language "ru" --model large-v2 --align_model "WAV2VEC2_ASR_LARGE_LV60K_960H" --task translate --max_line_count 2 --max_line_width 42 --highlight_words True
On Sat, 20 May 2023, 20:05 Simon, @.***> wrote:
Did you use --no_align?
— Reply to this email directly, view it on GitHub https://github.com/m-bain/whisperX/issues/267#issuecomment-1555918843, or unsubscribe https://github.com/notifications/unsubscribe-auth/A4MMREBCY7VHEGD3RAAWQL3XHDFSHANCNFSM6AAAAAAYIXZRWM . You are receiving this because you authored the thread.Message ID: @.***>
I think I got it. What's happening is that translate
can't align as of now (which is not documented I think) and as such we set the no_align
boolean to true
which itself is incompatible with highlight_words
(that's why I asked about it earlier). As such, highlight_words
can only be used with transcribe
right now. You sadly have to use transcribe and translate it afterwards.
Thanks for your reply @sorgfresser I think whisperx has translate issues.. hopefully it will be solved in future..
Can highlight_words somehow be used with python?
Whisperx can not highlight_words during translate.. it requires word_timestamps True..
When i input word_timestamps True then it does not work at all.. it said whisperx: error: unrecognized arguments: --word_timestamps True