An unofficial PyTorch implementation of VALL-E

Go to file

enhuiz f962a9d744 Fix raise		2023-01-13 00:34:40 +08:00
config	AdaLN and sample-wise quant level sampling	2023-01-13 00:33:16 +08:00
data/test	AdaLN and sample-wise quant level sampling	2023-01-13 00:33:16 +08:00
scripts	Make more VRAM friendly	2023-01-12 20:07:44 +08:00
vall_e	Fix raise	2023-01-13 00:34:40 +08:00
.gitignore	Python version	2023-01-12 20:37:04 +08:00
.gitmodules	Init	2023-01-11 23:03:25 +08:00
LICENSE	Create LICENSE	2023-01-12 15:01:26 +08:00
README.md	AdaLN and sample-wise quant level sampling	2023-01-13 00:33:16 +08:00
requirements.txt	Update docs	2023-01-12 14:56:30 +08:00
setup.py	Update setup.py	2023-01-12 20:45:59 +08:00

VALL-E

An unofficial PyTorch implementation of VALL-E, based on the EnCodec tokenizer.

Install

pip install git+https://github.com/enhuiz/vall-e

Note that the code is only tested under Python 3.10.7.

Put your data into a folder, e.g. data/your_data. Audio files should be named with the suffix .wav and text files with .normalized.txt.
Quantize the data:

python -m vall_e.emb.qnt data/your_data

python -m vall_e.emb.g2p data/your_data

Customize your configuration by creating config/your_data/ar.yml and config/your_data/nar.yml. Refer to the example configs in config/test and vall_e/config.py for details. You may choose different model presets, check vall_e/vall_e/__init__.py.
Train the AR or NAR model using the following scripts:

python -m vall_e.train yaml=config/your_data/ar_or_nar.yml