RVC-Boss

Results 484 comments of RVC-Boss

Pitch extract and one-click training error: Kernel size can't be greater than actual input size

> Is it supported on Apple devices? support m1/m2

__array_function__ internals

OSError: Model file not found: hubert_base.pt Maybe you need rename it to hubert_base.pth

ImportError: DLL load failed while importing _ufuncs: %1 不是有效的 Win32 应用程序。

系统是啥？

use default_factory

maybe it's because the version of fairseq Our repo is tested in fairseq==0.12.2

生成的歌声有很多电音和沙哑的声音

看不见训练数据，我猜想 1、前面有步骤报错，导致只训练了静音区，训练速度极快 2、生成的输入源/训练集未去伴奏处理

生成的歌声有很多电音和沙哑的声音

> 我觉得可能是训练集里有一些和声没去除干净。 去和声可以试试webui包里的去伴奏-HP5模型

有考虑过webui用tensorRT推理，手机端用ncnn推理吗？

大佬可以来贡献代码的呢 确实有这个需求

CUDA out of memory

If you GPU memory is 4G~6G, you should decrease batch size If you GPU memory is

[Feature request] Segmentation of audio based on parts to improve vocal output of long songs

Good idea.

[Feature request] Segmentation of audio based on parts to improve vocal output of long songs

'advanced settings' good idea