jiaheillu's picture
Upload 7 files
1745b9b

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: