how do i use train for single voice

anonymous-pits / pits

PITS: Variational Pitch Inference for End-to-end Pitch-controllable TTS without External Pitch Predictor

https://anonymous-pits.github.io/pits/

MIT License

274 stars 34 forks source link

how do i use train for single voice #15

Open wac81 opened 1 year ago

wac81 commented 1 year ago

how do i use train for single voice

anonymous-pits commented 1 year ago

You just need to change configs/config_en.yaml's speakers to single element list as ["spk0"].

wac81 commented 1 year ago

yes, i change this place ,and change sr to 22050,but get some error in inference for few sentences. most of sentences is get vocie. i don't know why?

torch shape is not match.

anonymous-pits commented 1 year ago

If torch shape is not matched, it seems that you modified segment_size: 8192. While its architecture from Avocodo, and it does not allow segment size change for fSBD.

Please give your full modification of config file to diagnosis your problems.