• This project targets deep learning beginners, basic knowledge of Python and PyTorch are the prerequisites for this project;
• This project aims to help deep learning beginners get rid of boring pure theoretical learning, and master the basic knowledge of deep learning by combining it with practices;
• This project does not support real-time voice converting; (need to replace whisper if real-time voice converting is what you are looking for)
• This project will not develop one-click packages for other purposes;

• A minimum VRAM requirement of 6GB for training

• Support for multiple speakers

• Create unique speakers through speaker mixing

• It can even convert voices with light accompaniment

• You can edit F0 using Excel

Powered by @ShadowVap

due to the use of data perturbation, it takes longer to train than other projects.

USP : Unvoice and Silence with Pitch when infer

1. Install PyTorch.

2. Install project dependencies

   pip install -i https://pypi.tuna.tsinghua.edu.cn/simple -r requirements.txt

   Note: whisper is already built-in, do not install it again otherwise it will cuase conflict and error

3. Download the Timbre Encoder: Speaker-Encoder by @mueller91, put best_model.pth.tar into speaker_pretrain/.

4. Download whisper model whisper-large-v2. Make sure to download large-v2.pt，put it into whisper_pretrain/.

5. Download hubert_soft model，put hubert-soft-0d54a1f4.pt into hubert_pretrain/.

6. Download pitch extractor crepe full，put full.pth into crepe/assets.

   Note: crepe full.pth is 84.9 MB, not 6kb

7. Download pretrain model sovits5.0.pretrain.pth, and put it into vits_pretrain/.

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

Necessary pre-processing:

1. Separate voice and accompaniment with UVR (skip if no accompaniment)
2. Cut audio input to shorter length with slicer, whisper takes input less than 30 seconds.
3. Manually check generated audio input, remove inputs shorter than 2 seconds or with obivous noise.
4. Adjust loudness if necessary, recommend Adobe Audiiton.
5. Put the dataset into the dataset_raw directory following the structure below.

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

python svc_preprocessing.py -t 2

-t: threading, max number should not exceed CPU core count, usually 2 is enough. After preprocessing you will get an output with following structure.

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
└── hubert
│    └── speaker0
│    │      ├── 000001.vec.npy
│    │      └── 000xxx.vec.npy
│    └── speaker1
│           ├── 000001.vec.npy
│           └── 000xxx.vec.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
|
└── indexes
    ├── speaker0
    │   ├── some_prefix_hubert.index
    │   └── some_prefix_whisper.index
    └── speaker1
        ├── hubert.index
        └── whisper.index

1. Re-sampling

   • Generate audio with a sampling rate of 16000Hz in ./data_svc/waves-16k

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

   • Generate audio with a sampling rate of 32000Hz in ./data_svc/waves-32k

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

2. Use 16K audio to extract pitch

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

3. Use 16K audio to extract ppg

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

4. Use 16K audio to extract hubert

   python prepare/preprocess_hubert.py -w data_svc/waves-16k/ -v data_svc/hubert
   

5. Use 16k audio to extract timbre code

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

6. Extract the average value of the timbre code for inference; it can also replace a single audio timbre in generating the training index, and use it as the unified timbre of the speaker for training

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

7. Use 32k audio to extract the linear spectrum

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

8. Use 32k audio to generate training index

   python prepare/preprocess_train.py
   

9. Training file debugging

   python prepare/preprocess_zzz.py
   

1. If fine-tuning is based on the pre-trained model, you need to download the pre-trained model: sovits5.0.pretrain.pth. Put pretrained model under project root, change this line

   pretrain: "./vits_pretrain/sovits5.0.pretrain.pth"
   

   in configs/base.yaml，and adjust the learning rate appropriately, eg 5e-5.

   batch_size: for GPU with 6G VRAM, 6 is the recommended value, 8 will work but step speed will be much slower.

2. Start training

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

3. Resume training

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

4. Log visualization

   tensorboard --logdir logs/
   

1. Export inference model: text encoder, Flow network, Decoder network

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

2. Inference

   • if there is no need to adjust f0, just run the following command.

   python svc_inference.py --config configs/base.yaml --model sovits5.0.pth --spk ./data_svc/singer/your_singer.spk.npy --wave test.wav --shift 0
   

   • if f0 will be adjusted manually, follow the steps:
     1. use whisper to extract content encoding, generate test.vec.npy.

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

     2. use hubert to extract content vector, without using one-click reasoning, in order to reduce GPU memory usage

     python hubert/inference.py -w test.wav -v test.vec.npy
     

     3. extract the F0 parameter to the csv text format, open the csv file in Excel, and manually modify the wrong F0 according to Audition or SonicVisualiser

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

     4. final inference

     python svc_inference.py --config configs/base.yaml --model sovits5.0.pth --spk ./data_svc/singer/your_singer.spk.npy --wave test.wav --ppg test.ppg.npy --vec test.vec.npy --pit test.csv --shift 0
     

3. Notes

   • when --ppg is specified, when the same audio is reasoned multiple times, it can avoid repeated extraction of audio content codes; if it is not specified, it will be automatically extracted;

   • when --vec is specified, when the same audio is reasoned multiple times, it can avoid repeated extraction of audio content codes; if it is not specified, it will be automatically extracted;

   • when --pit is specified, the manually tuned F0 parameter can be loaded; if not specified, it will be automatically extracted;

   • generate files in the current directory:svc_out.wav

4. Arguments ref

   args	--config	--model	--spk	--wave	--ppg	--vec	--pit	--shift
   name	config path	model path	speaker	wave input	wave ppg	wave hubert	wave pitch	pitch shift

5. post by vad

python svc_inference_post.py --ref test.wav --svc svc_out.wav --out svc_out_post.wav

To increase the stability of the generated timbre, you can use the method described in the Retrieval-based-Voice-Conversion repository. This method consists of 2 steps:

1. Training the retrieval index on hubert and whisper features Run training with default settings:

   python svc_train_retrieval.py
   

   If the number of vectors is more than 200_000 they will be compressed to 10_000 using the MiniBatchKMeans algorithm. You can change these settings using command line options:

   usage: crate faiss indexes for feature retrieval [-h] [--debug] [--prefix PREFIX] [--speakers SPEAKERS [SPEAKERS ...]] [--compress-features-after COMPRESS_FEATURES_AFTER]
                                                    [--n-clusters N_CLUSTERS] [--n-parallel N_PARALLEL]
   
   options:
     -h, --help            show this help message and exit
     --debug
     --prefix PREFIX       add prefix to index filename
     --speakers SPEAKERS [SPEAKERS ...]
                           speaker names to create an index. By default all speakers are from data_svc
     --compress-features-after COMPRESS_FEATURES_AFTER
                           If the number of features is greater than the value compress feature vectors using MiniBatchKMeans.
     --n-clusters N_CLUSTERS
                           Number of centroids to which features will be compressed
     --n-parallel N_PARALLEL
                           Nuber of parallel job of MinibatchKmeans. Default is cpus-1
   

   Compression of training vectors can speed up index inference, but reduces the quality of the retrieve. Use vector count compression if you really have a lot of them.

   The resulting indexes will be stored in the "indexes" folder as:

   data_svc
   ...
   └── indexes
       ├── speaker0
       │   ├── some_prefix_hubert.index
       │   └── some_prefix_whisper.index
       └── speaker1
           ├── hubert.index
           └── whisper.index
   

2. At the inference stage adding the n closest features in a certain proportion of the vits model Enable Feature Retrieval with settings:

   python svc_inference.py --config configs/base.yaml --model sovits5.0.pth --spk ./data_svc/singer/your_singer.spk.npy --wave test.wav --shift 0 \
   --enable-retrieval \
   --retrieval-ratio 0.5 \
   --n-retrieval-vectors 3
   

   For a better retrieval effect, you can try to cycle through different parameters: --retrieval-ratio and --n-retrieval-vectors

   If you have multiple sets of indexes, you can specify a specific set via the parameter: --retrieval-index-prefix

   You can explicitly specify the paths to the hubert and whisper indexes using the parameters: --hubert-index-path and --whisper-index-path

named by pure coincidence：average -> ave -> eva，eve(eva) represents conception and reproduction

python svc_eva.py

eva_conf = {
    './configs/singers/singer0022.npy': 0,
    './configs/singers/singer0030.npy': 0,
    './configs/singers/singer0047.npy': 0.5,
    './configs/singers/singer0051.npy': 0.5,
}

the generated singer file will be eva.spk.npy.

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/huawei-noah/Speech-Backbones/tree/main/Grad-TTS

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

https://github.com/mozilla/TTS

https://github.com/bshall/soft-vc

https://github.com/maxrmorrison/torchcrepe

https://github.com/MoonInTheRiver/DiffSinger

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

https://github.com/yl4579/HiFTNet paper

Autoregressive neural f0 model for statistical parametric speech synthesis

One-shot Voice Conversion by Separating Speaker and Content Representations with Instance Normalization

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

AdaVITS: Tiny VITS for Low Computing Resource Speaker Adaptation

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

Multilingual Speech Synthesis and Cross-Language Voice Cloning: GRL

RoFormer: Enhanced Transformer with rotary position embedding

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

https://github.com/Francis-Komizu/Sovits

• LoRA-SVC: decoder only svc
• Grad-SVC: diffusion based svc

2022.04.12 https://mp.weixin.qq.com/s/autNBYCsG4_SvWt2-Ll_zA

2022.04.22 https://github.com/PlayVoice/VI-SVS

2022.07.26 https://mp.weixin.qq.com/s/qC4TJy-4EVdbpvK2cQb1TA

2022.09.08 https://github.com/PlayVoice/VI-SVC