I am using the voxtral transcribe endpoint, which currently only offers voxtral-mini (rather than the prbably more accurate voxtral-small). I am finding many of my transcripts have random sections that are simply dropped, sometimes concatenating parts of two sentences together because the end of one and the beginning of the next were skipped. Listening to the sections of the audio in question, I rarely find any audio quality issue that might have lead to this. The only reason I can immediately think of is if silence segments are being cut out in pre-processing, but accidentally cutting surrounding audio segments at the same time.
#Voxtral Mini Transcribe drops sections of audio.
3 messages · Page 1 of 1 (latest)
Thanks for the Feedback!
Thanks a lot for the feedback! If you need help in the future, you can contact support to get the help you need, visit the following article to learn how to contact support.
'Resolve' will close this feedback post.
What I experience is cutting off segments at the end of the transcription often, at least in German. See my separate feedback. Seems others experience that too.