search

NVIDIA / waveglow

A Flow-based Generative Network for Speech Synthesis
BSD 3-Clause "New" or "Revised" License
2.27k stars 531 forks source link

Waveglow on multi speaker #219

Open sbkim052 opened 4 years ago

sbkim052 commented 4 years ago

Have anyone succeeded on using Waveglow for multispeaker vocoder?

rafaelvalle commented 4 years ago

The samples provided at https://nv-adlr.github.io/Flowtron and https://nv-adlr.github.io/Mellotron were generated with a single WaveGlow trained on a single speaker.

hongchengzhu commented 2 years ago

Have anyone succeeded on using Waveglow for multispeaker vocoder?

Hello, have you tried to train and infer in multispeaker setting? Does it work? How is the inference quality? Thanks.