# 绫华语音模型训练参数



duration of dataset：72min 

epoch: 3500

batch_size :24

segments = 452

step:32800

Losses vs step：

![](./README.assets/%E6%88%AA%E5%B1%8F2023-04-05%2018.38.00.png)