Datasets:

jiaheillu
/

sovits_audio_preview

# 宵宫语音模型训练参数



duration of dataset：59min 

epoch: 3500

batch_size :24

segments = 339

step:36800

Losses vs step

![截屏2023-04-07 17.39.57](./README.assets/%E6%88%AA%E5%B1%8F2023-04-07%2017.39.57.png)