I was think about the tensorboard, i set the smoothing to 0.987 then the graph increasing for loss/g/total for overtraining and i use the UVR-Denoise for the first time.

Sample Rate: 40k

Dataset: 15 Minutes (All Songs)

Batch Size 4

Credit to me Youtube & Tiktok

Index Ratio 1.0 Recommended

If you wanted to take down for low quality, I'll remade dataset or maybe i will move on to the next one.

https://huggingface.co/CrimsonZockt/RVC-Model/resolve/main/Teyadora.zip

Additional Details:

Tags: No tags available

Download Link: https://huggingface.co/CrimsonZockt/RVC-Model/resolve/main/Teyadora.zip