search

KimythAnly / AGAIN-VC

This is the official implementation of the paper AGAIN-VC: A One-shot Voice Conversion using Activation Guidance and Adaptive Instance Normalization.
https://kimythanly.github.io/AGAIN-VC-demo/index
MIT License
111 stars 19 forks source link

How to Use Self built Datasets #12

Closed zerclanzhang closed 3 months ago

zerclanzhang commented 2 years ago

Hello dude, the code could run well with CYTK datastes, but when i use the self-built dataset which contains only 2 speakers, error showed as : train.py no dev_iter.

Any tips on how to use self built datasets?

So far, i can only train without evaluation by setting eva_steps larger than total_steps.

zerclanzhang commented 2 years ago

part of my dataset was extracted from aidatang2000, which is 16000 sample rate chinese dataset. I've already changed all the related sample rates to 16000, and f_max: 8000 instead in preprocess config.