VITS TTS Model Tutorial with David Attenborough AI #artificialintelligence #ai #tts #coqui

Описание

youmebangbang@gmail.com
Music: Com Truise: Iteration

MODEL LINKS:
https://huggingface.co/youmebangbang/vits_tts_models

SETUP AND TRAINING:
git clone https://github.com/coqui-ai/TTS.git
cd TTS
pip install -e .[all, notebooks]
navigate to TTS/recipes/ljspeech/vits
create a project folder and copy wav folder and metadata.csv to it
edit the train_vits.py file
python train_vits.py

INFERENCE:
pip install TTS

For Windows only:
install espeak: https://espeak.sourceforge.net/download.html
pip install mecab
If you get an error about mecab then download mecab-0.996.exe (not the DLL as video says) from https://drive.google.com/uc?export=download&id=0B4y35FiV1wh7WElGUGt6ejlpVXc and place it in your mecab site packages folder

Windows & Linux:
TTS --text "your text" --model_path da_checkpoint.pth --config_path config.json --out_path out.wav --use_cuda USE_CUDA
Github fork of synthesizer.py to join short sentences and separate wavs: https://github.com/youmebangbang/TTS/blob/dev/TTS/utils/synthesizer.py