VISinger2

🍴 Forked from zhangyongmao/VISinger2

📌 This clone repo is created as GitHub does not support Git LFS on public fork

Fixed PyTorch 2.0.1 compatibility issues, now VISinger 2 can be run in cuda12.

Updates

• Add FastAPI server for serving inference
• Added Cantonese singing data for training
• Added Cantonese pre-trained model

Pre-requisites

1. Install python requirements: pip install -r requirements.txt
2. Download the Opencpop Dataset.
3. prepare data like data/opencpop (wavs, train.txt, test.txt, train.list, test.list)
4. To generate train.list, use awk -F'|' '{print $1}' train.txt > train.list
5. To generate test.list, use awk -F'|' '{print $1}' test.txt > test.list
6. modify the egs/visinger2/config.json (data/data_dir, train/save_dir)

extract pitch and mel

cd egs/visinger2
bash bash/preprocess.sh config.json

Training

cd egs/visinger2
bash bash/train.sh 0

We trained the model for 500k steps with batch size of 16.

Inference

modify the model_dir, input_dir, output_dir in inference.sh

cd egs/visinger2
bash bash/inference.sh

Some audio samples can be found in demo website and bilibili.

The pre-trained model trained using opencpop is here, the config.json is here, and the result of the test set synthesized by this pre-trained model is here.