What sampling rate was the model trained on?

wenet-e2e / wespeaker

Research and Production Oriented Speaker Verification, Recognition and Diarization Toolkit

Apache License 2.0

630 stars 109 forks source link

What sampling rate was the model trained on? #244

Closed youssefabdelm closed 7 months ago

youssefabdelm commented 7 months ago

Trying to run it on many files that are 44.1kHz and not sure if I should downsample to 16kHz before running it through the model or not?

wsstriving commented 7 months ago

Yes, you need to downsample it to 16kHz, all the pretrained models currently are trained on 16k data

youssefabdelm commented 7 months ago

Thanks so much! And mono as well?

JiJiJiang commented 7 months ago

Yes, we support mono only. BTW, you can set resample_rate as 16k in the config yaml file and resampling would be done during the training process, instead of downsampling all wavs into 16kHZ before training.

wsstriving commented 7 months ago

Has been solved