Jun Chen

Results 17 comments of Jun Chen

Floating point exception (core dumped), but can't find core dump in /var/lib/systemd/coredump/

> How did you get this error? > > Can you share the steps to reproduce? (model size, transcription options, input file if possible) Hi, sorry I can't share the...

Floating point exception (core dumped), but can't find core dump in /var/lib/systemd/coredump/

I basically tried the notebook here: https://github.com/Majdoddin/nlp/blob/main/Pyannote_plays_and_Whisper_rhymes_v_2_0.ipynb It's not happening on every of my input file though

Floating point exception (core dumped), but can't find core dump in /var/lib/systemd/coredump/

thanks, I'll try today or tomorrow

Floating point exception (core dumped), but can't find core dump in /var/lib/systemd/coredump/

sorry was busy with some deadlines. I wasn't able to run that library to repro today after tried for 10+mins, but since I'll not use it, I assume this is...

Timestamp of first word after long silence in not accurate sometimes

I think encountered this issue as well, I split the stereo audio to two mono channels, there're expected long silence in each mono audio. Using vad_filter=True fixed a few, but...

Timestamp of first word after long silence in not accurate sometimes

Thanks @guillaumekln ! Yeah in my experience end is more accurate than start. So accounting for end will be more accurate. I have one question about `speech_pad_ms` Does it just...

Timestamp of first word after long silence in not accurate sometimes

> > So as long as it's smaller than min_speech_duration_ms, then should be fine? > > Did you mean `min_silence_duration_ms`? sorry yes I mean `min_silence_duration_ms` =.= Actually I have another...

timestamp not matching well when run transcribe on two mono audios split from stereo and assemble back

Will try with predict speaker and not split

timestamp not matching well when run transcribe on two mono audios split from stereo and assemble back

that doesn't work well, I probably still need to split and improve the timestamps matchings -.-

This is very cool, but push to even higher gpu usage?

> > Tried to use a thread pool in python to submit jobs for audios, > > That's a good approach. Did you also increase `num_workers` when doing that? Normally...