"Uma Musume" Mambo/Poetic Opera GPT-SoVITS model

Looking at GPT-SoVITS, which doesn't seem to be very popular yet, today I created a model.

This model is trained based on the V2 version. The V1 version does not guarantee usability; it is best to use the V2 version.

When using, please place the model files in the corresponding folders. The RVC version of the model is already in training, including support for Chinese and Japanese.

Short videos and the popular poetry and opera series on Bilibili are produced using this timbre.

The folder includes

:

Model files

A training dataset and a test speech dataset

Model audition

<!--[if lt IE 9]>document.createElement('audio');<![endif]-->

Screenshot of alchemy

All the raw datasets above are manually extracted and produced, so the results are better. However, since these are daily conversions, the sound may sound a bit awkward. This is similar to gacha—just draw a few more times, or you can select test datasets from the raw datasets we provide for inference.