search

hitachi-speech / EEND

End-to-End Neural Diarization
MIT License
368 stars 57 forks source link

Confusion about the store of uttid #9

Closed 875441459 closed 3 years ago

875441459 commented 4 years ago

hi, I'm a little confused about the calculation and store of uttid. In makemixture.py, when calculating the uttid, it multiples 100, like uttid = '{}{}{:07d}{:07d}'.format(spkid, recid, int(startpos / args.rate 100),int(endpos / args.rate 100)).But why not do sth like str(int(startpos / args.rate *1000)/1000) to store it more precisely?

yubouf commented 4 years ago

We followed the kaldi-style. https://kaldi-asr.org/doc/data_prep.html Multiply by 100 means the frame index when using 10 msec frame shift. I think this resolution is sufficient for our tasks.

875441459 commented 4 years ago

Thanks, I'll check the website.