search

aalto-speech / speaker-diarization

Speaker diarization scripts, based on AaltoASR
190 stars 37 forks source link

WAV files format #9

Open RefiPeretz opened 7 years ago

RefiPeretz commented 7 years ago

When using your wav example when using spk-diarization2.py everything works pretty good. However when I try to use on my WAV files I get strange results. I try to compare WAV properties between my files and yours: Input #0, wav, from 'meeting.wav': Duration: 00:09:25.60, bitrate: 256 kb/s Stream #0:0: Audio: pcm_s16le ([1][0][0][0] / 0x0001), 16000 Hz, 1 channels, s16, 256 kb/s I manipulated my wav file to be with the same properties using ffmpeg but still it doesn't work.

I cant understand what more can I do. Is there any manipulation I need to do to the WAV files before using the script?

RefiPeretz commented 7 years ago

I added a wav example where the script performance is really poor. This file is before any manipulation. Maya.wav.zip

antoniomo commented 7 years ago

Hi,

I would need to test it to have an answer, sorry it takes me a few days as I'm currently busy :)

RefiPeretz commented 7 years ago

Sure. It so strange I already tried with several files, nothing works except meeting.wav. Waiting to hear from you.

antoniomo commented 7 years ago

No worries I'll take a look! In the meantime check this (maybe related, maybe not) issue: https://github.com/aalto-speech/speaker-diarization/issues/7