search

clovaai / voxceleb_trainer

In defence of metric learning for speaker recognition
MIT License
1.02k stars 272 forks source link

Use as Speaker Encoder for 48kHz audio #138

Closed astricks closed 2 years ago

astricks commented 2 years ago

Hello,

I am interested in a 48kHz version of this model. The provided model works very well with 22kHz audio as a speaker encoder for voice cloning TTS models. I now want to try it with 48kHz audio.

I just wanted to check if there was a model already available that was trained on a 48kHz dataset. Appreciate your time and help.

Warm regards.

astricks commented 2 years ago

Never mind, I figured it out.

predawnang commented 1 year ago

Hi, do you mean pre-trained rawnet3 model works very well with 22kHz audio? I recently plan to use rawnet3 within my project, but i'm not sure if the pre-trained version of rawnet3 works fine with 22050 sample rate raw waveform.