iTA9178 / so-vits-svc-5.0

Core Engine of Singing Voice Conversion & Singing Voice Clone

Home Page:https://huggingface.co/spaces/maxmax20160403/sovits5.0

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Variational Inference with adversarial learning for end-to-end Singing Voice Conversion based on VITS

Hugging Face Spaces Open in Colab GitHub Repo stars GitHub forks GitHub issues GitHub

  • 💗本项目的目标群体是:深度学习初学者,具备Python和PyTorch的基本操作是使用本项目的前置条件;
  • 💗本项目旨在帮助深度学习初学者,摆脱枯燥的纯理论学习,通过与实践结合,熟练掌握深度学习基本知识;
  • 💗本项目不支持实时变声;(也许以后会支持,但要替换掉whisper)
  • 💗本项目不会开发用于其他用途的一键包。(不会指没学会)

sovits_framework

  • 【低 配置】6G显存可训练

  • 【无 泄漏】支持多发音人

  • 【带 伴奏】也能进行转换,轻度伴奏

  • 【用 Excel】进行原始调教,纯手工

本项目并不基于svc-develop-team/so-vits-svc,恰恰相反,见https://github.com/svc-develop-team/so-vits-svc/tree/2.0

本项目将继续完成基于BIGVGAN的模型(32K),在此之后,有成果再更新项目

  • 5.0.epoch1200.full.pth模型包括:生成器+判别器=176M,可用作预训练模型
  • 发音人(56个)文件在configs/singers目录中,可进行推理测试,尤其测试音色泄露
  • 发音人22,30,47,51辨识度较高,音频样本在configs/singers_sample目录中
Feature From Status Function Remarks
whisper OpenAI 强大的抗噪能力 参数修改
bigvgan NVIDA 抗锯齿与蛇形激活 GPU占用略多,主分支删除;新分支训练,共振峰更清晰,提升音质明显
natural speech Microsoft 减少发音错误 -
neural source-filter NII 解决断音问题 参数优化
speaker encoder Google 音色编码与聚类 -
GRL for speaker Ubisoft 防止编码器泄露音色 原理类似判别器的对抗训练
one shot vits Samsung VITS 一句话克隆 -
SCLN Microsoft 改善克隆 -
band extention Adobe 16K升48K采样 数据处理
PPG perturbation 本项目 提升抗噪性和去音色 -

💗GRL去音色泄漏,更多的是理论上的价值;Hugging Face Demo推理模型无泄漏主要归因于PPG扰动;由于使用了数据扰动,相比其他项目需要更长的训练时间。

数据集准备

uvr5_config

💗必要的前处理:

然后以下面文件结构将数据集放入dataset_raw目录

dataset_raw
├───speaker0
│   ├───000001.wav
│   ├───...
│   └───000xxx.wav
└───speaker1
    ├───000001.wav
    ├───...
    └───000xxx.wav

安装依赖

数据预处理

  • 1, 设置工作目录:heartpulse::heartpulse::heartpulse:不设置后面会报错

    export PYTHONPATH=$PWD

  • 2, 重采样

    生成采样率16000Hz音频, 存储路径为:./data_svc/waves-16k

    python prepare/preprocess_a.py -w ./data_raw -o ./data_svc/waves-16k -s 16000

    生成采样率32000Hz音频, 存储路径为:./data_svc/waves-32k

    python prepare/preprocess_a.py -w ./data_raw -o ./data_svc/waves-32k -s 32000

    可选的16000Hz提升到32000Hz,待完善~批处理

    python bandex/inference.py -w svc_out.wav

  • 3, 使用16K音频,提取音高

    python prepare/preprocess_f0.py -w data_svc/waves-16k/ -p data_svc/pitch

  • 4, 使用16k音频,提取内容编码

    python prepare/preprocess_ppg.py -w data_svc/waves-16k/ -p data_svc/whisper

  • 5, 使用16k音频,提取音色编码;应该将speaker改为timbre,才准确

    python prepare/preprocess_speaker.py data_svc/waves-16k/ data_svc/speaker

  • 6, 提取音色编码均值,用于推理;也可以在生成训练索引中,替换单个音频音色,作为发音人统一音色用于训练

    python prepare/preprocess_speaker_ave.py data_svc/speaker/ data_svc/singer

  • 7, 使用32k音频,提取线性谱

    python prepare/preprocess_spec.py -w data_svc/waves-32k/ -s data_svc/specs

  • 8, 使用32k音频,生成训练索引

    python prepare/preprocess_train.py

  • 9, 训练文件调试

    python prepare/preprocess_zzz.py

data_svc/
│
└── waves-16k
│    │
│    └── speaker0
│    │      ├── 000001.wav
│    │      └── 000xxx.wav
│    └── speaker1
│           ├── 000001.wav
│           └── 000xxx.wav
│
└── waves-32k
│    │
│    └── speaker0
│    │      ├── 000001.wav
│    │      └── 000xxx.wav
│    └── speaker1
│           ├── 000001.wav
│           └── 000xxx.wav
│
└── pitch
│    │
│    └── speaker0
│    │      ├── 000001.pit.npy
│    │      └── 000xxx.pit.npy
│    └── speaker1
│           ├── 000001.pit.npy
│           └── 000xxx.pit.npy
│
└── whisper
│    │
│    └── speaker0
│    │      ├── 000001.ppg.npy
│    │      └── 000xxx.ppg.npy
│    └── speaker1
│           ├── 000001.ppg.npy
│           └── 000xxx.ppg.npy
│
└── speaker
│    │
│    └── speaker0
│    │      ├── 000001.spk.npy
│    │      └── 000xxx.spk.npy
│    └── speaker1
│           ├── 000001.spk.npy
│           └── 000xxx.spk.npy
|
└── singer
    ├── speaker0.spk.npy
    └── speaker1.spk.npy

训练

  • 0, 如果基于预训练模型微调,需要下载预训练模型5.0.epoch1200.full.pth

    指定configs/base.yaml参数pretrain: "",并适当调小学习率

  • 1, 设置工作目录:heartpulse::heartpulse::heartpulse:不设置后面会报错

    export PYTHONPATH=$PWD

  • 2, 启动训练

    python svc_trainer.py -c configs/base.yaml -n sovits5.0

  • 3, 恢复训练

    python svc_trainer.py -c configs/base.yaml -n sovits5.0 -p chkpt/sovits5.0/***.pth

  • 4, 查看日志,release页面有完整的训练日志

    tensorboard --logdir logs/

20K一阶段训练日志如下,可以看到还未收敛完成

sovits5 0 preview

sovits_spec

推理

  • 1, 设置工作目录:heartpulse::heartpulse::heartpulse:不设置后面会报错

    export PYTHONPATH=$PWD

  • 2, 导出推理模型:文本编码器,Flow网络,Decoder网络;判别器和后验编码器只在训练中使用

    python svc_export.py --config configs/base.yaml --checkpoint_path chkpt/sovits5.0/***.pt

  • 3, 使用whisper提取内容编码,没有采用一键推理,为了降低显存占用

    python whisper/inference.py -w test.wav -p test.ppg.npy

    生成test.ppg.npy;如果下一步没有指定ppg文件,则调用程序自动生成

  • 4, 提取csv文本格式F0参数,Excel打开csv文件,对照Audition或者SonicVisualiser手动修改错误的F0

    python pitch/inference.py -w test.wav -p test.csv

sonic visualiser

  • 5,指定参数,推理

    python svc_inference.py --config configs/base.yaml --model sovits5.0.pth --spk ./configs/singers/singer0001.npy --wave test.wav --ppg test.ppg.npy --pit test.csv

    当指定--ppg后,多次推理同一个音频时,可以避免重复提取音频内容编码;没有指定,也会自动提取;

    当指定--pit后,可以加载手工调教的F0参数;没有指定,也会自动提取;

    生成文件在当前目录svc_out.wav;

    args --config --model --spk --wave --ppg --pit
    name 配置文件 模型文件 音色文件 音频文件 音频内容 音高内容

数据集

Name URL
KiSing http://shijt.site/index.php/2021/05/16/kising-the-first-open-source-mandarin-singing-voice-synthesis-corpus/
PopCS https://github.com/MoonInTheRiver/DiffSinger/blob/master/resources/apply_form.md
opencpop https://wenet.org.cn/opencpop/download/
Multi-Singer https://github.com/Multi-Singer/Multi-Singer.github.io
M4Singer https://github.com/M4Singer/M4Singer/blob/master/apply_form.md
CSD https://zenodo.org/record/4785016#.YxqrTbaOMU4
KSS https://www.kaggle.com/datasets/bryanpark/korean-single-speaker-speech-dataset
JVS MuSic https://sites.google.com/site/shinnosuketakamichi/research-topics/jvs_music
PJS https://sites.google.com/site/shinnosuketakamichi/research-topics/pjs_corpus
JUST Song https://sites.google.com/site/shinnosuketakamichi/publication/jsut-song
MUSDB18 https://sigsep.github.io/datasets/musdb.html#musdb18-compressed-stems
DSD100 https://sigsep.github.io/datasets/dsd100.html
Aishell-3 http://www.aishelltech.com/aishell_3
VCTK https://datashare.ed.ac.uk/handle/10283/2651

代码来源和参考文献

https://github.com/facebookresearch/speech-resynthesis paper

https://github.com/jaywalnut310/vits paper

https://github.com/openai/whisper/ paper

https://github.com/NVIDIA/BigVGAN paper

https://github.com/mindslab-ai/univnet paper

https://github.com/nii-yamagishilab/project-NN-Pytorch-scripts/tree/master/project/01-nsf

https://github.com/brentspell/hifi-gan-bwe

https://github.com/mozilla/TTS

https://github.com/OlaWod/FreeVC paper

SNAC : Speaker-normalized Affine Coupling Layer in Flow-based Architecture for Zero-Shot Multi-Speaker Text-to-Speech

Adapter-Based Extension of Multi-Speaker Text-to-Speech Model for New Speakers

AdaSpeech: Adaptive Text to Speech for Custom Voice

Cross-Speaker Prosody Transfer on Any Text for Expressive Speech Synthesis

Learn to Sing by Listening: Building Controllable Virtual Singer by Unsupervised Learning from Voice Recordings

Adversarial Speaker Disentanglement Using Unannotated External Data for Self-supervised Representation Based Voice Conversion

Speaker normalization (GRL) for self-supervised speech emotion recognition

基于数据扰动防止音色泄露的方法

https://github.com/auspicious3000/contentvec/blob/main/contentvec/data/audio/audio_utils_1.py

https://github.com/revsic/torch-nansy/blob/main/utils/augment/praat.py

https://github.com/revsic/torch-nansy/blob/main/utils/augment/peq.py

https://github.com/biggytruck/SpeechSplit2/blob/main/utils.py

https://github.com/OlaWod/FreeVC/blob/main/preprocess_sr.py

贡献者

交流群

炼丹师公会-SVC群聊二维码

About

Core Engine of Singing Voice Conversion & Singing Voice Clone

https://huggingface.co/spaces/maxmax20160403/sovits5.0

License:MIT License


Languages

Language:Python 100.0%