search

espnet / espnet

End-to-End Speech Processing Toolkit
https://espnet.github.io/espnet/
Apache License 2.0
8.39k stars 2.17k forks source link

egs for CUM Audio dataset #2943

Closed roholazandie closed 3 years ago

roholazandie commented 3 years ago

Hello, Is there any egs for CMU ARCTIC dataset in espnet?

http://www.speech.cs.cmu.edu/databases/an4/

Does this have acceptable quality for training the vocoder and TTS system for male voice?

kan-bayashi commented 3 years ago

You mean CMU arctic database? http://www.festvox.org/cmu_arctic/ Arctic dataset is a little bit small to train E2E-TTS model from scratch. egs/arctic/tts1 is the example to perform fine-tuning but I did not prepare for ESPnet2.

For the vocoder, PWG is fine (MelGAN is not stable due to the small amount of data). You can find egs in my repo.