Skip to content

Commit 59a987e

Browse files
committed
update readmes
1 parent 8c54218 commit 59a987e

File tree

2 files changed

+4
-4
lines changed

2 files changed

+4
-4
lines changed

README.md

Lines changed: 3 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -65,7 +65,7 @@ python [export_ckpt.py](export_ckpt.py) --ckpt_path ckpt路径 --save_path 导
6565

6666
因为 pytorch-lightning 的问题所以说在 GAN 训练过程中实际的步数是它显示步数的一半
6767

68-
如果你需要微调社区声码器权重建议使用[ft_hifigan.yaml](configs/ft_hifigan.yaml) 配置文件
68+
如果你需要微调社区声码器请使用[ft_hifigan.yaml](configs/ft_hifigan.yaml) 配置文件,并用 'finetune_ckpt_path' 选项指定权重路径
6969

7070
如何使用微调功能建议参考 openvpi/diffsinger [项目文档](https://github.com/openvpi/DiffSinger/blob/main/docs/BestPractices.md#fine-tuning-and-parameter-freezing)
7171

@@ -106,11 +106,11 @@ python export_ckpt.py --ckpt_path (your ckpt path) --save_path (output ckpt pat
106106
# 注意事项
107107
实际步数是显示的一半
108108

109-
微调 nsf-hifigan 声码器请将 [releases](https://github.com/openvpi/SingingVocoders/releases) 中的权重解压后放至主目录下,并使用 [ft_hifigan.yaml](configs/ft_hifigan.yaml)
109+
微调 nsf-hifigan 声码器请下载并解压 [releases](https://github.com/openvpi/SingingVocoders/releases) 中的权重,并将 [ft_hifigan.yaml](configs/ft_hifigan.yaml) 中的 'finetune_ckpt_path' 选项改为权重路径
110110

111111
微调请使用 44100 Hz 采样率音频,并不要修改其他 mel 参数,除非你明确知道你在做什么
112112

113-
微调功能使用请参考 openvpi/DiffSinger [项目文档](https://github.com/openvpi/DiffSinger/blob/main/docs/BestPractices.md#fine-tuning-and-parameter-freezing)
113+
微调的其他功能使用请参考 openvpi/DiffSinger [项目文档](https://github.com/openvpi/DiffSinger/blob/main/docs/BestPractices.md#fine-tuning-and-parameter-freezing)
114114

115115
导出的权重可以在 [DDSP-SVC](https://github.com/yxlllc/DDSP-SVC), [Diffusion-SVC](https://github.com/CNChTu/Diffusion-SVC), [so-vits-svc](https://github.com/svc-develop-team/so-vits-svc)[DiffSinger (openvpi)](https://github.com/openvpi/DiffSinger) 等项目中使用
116116

README_en.md

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -77,7 +77,7 @@ Note that data augmentation may damage the sound quality!
7777
# Note
7878
Because of some problems the actual number of steps is half of what it shows
7979
80-
To fine-tune the nsf-hifigan vocoder, please unzip the weights in [releases](https://github.com/openvpi/SingingVocoders/releases) and put them in the main directory, and use [ft_hifigan.yaml](configs%2Fft_hifigan.yaml)
80+
To fine-tune the nsf-hifigan vocoder, please download and unzip the weights in [releases](https://github.com/openvpi/SingingVocoders/releases), and modify the 'finetune_ckpt_path' item in [ft_hifigan.yaml](configs%2Fft_hifigan.yaml) to the checkpoint file.
8181
8282
For fine-tuning please use 44100 Hz samplerate audio and do not modify other mel parameters unless you know exactly what you are doing
8383

0 commit comments

Comments
 (0)