search

auspicious3000 / autovc

AutoVC: Zero-Shot Voice Style Transfer with Only Autoencoder Loss
https://arxiv.org/abs/1905.05879
MIT License
983 stars 207 forks source link

About the trained vocoder #59

Closed Baliii closed 3 years ago

Baliii commented 3 years ago

Hi, thanks for your great contributions. The WaveNet Vocoder you provide in Pre-trained models is based on which dataset? Can I train the vocoder on my dataset?

auspicious3000 commented 3 years ago

Please refer to paper section 4.4