RADTTS + WaveODE vocoder

• WaveODE: https://github.com/WelkinYang/WaveODE

🇺🇦 Join Ukrainian Text-to-Speech community: https://t.me/speech_synthesis_uk

---

Clone the code:

git clone https://github.com/egorsmkv/radtts-waveode
cd radtts-waveode

Install dependencies:

pip install -r requirements.txt

Download Ukrainian RADTTS and WaveODE models:

mkdir models
cd models

wget https://github.com/egorsmkv/radtts-waveode/releases/download/v1.0/M_105.pth
wget https://github.com/egorsmkv/radtts-istftnet/releases/download/v1.0/RADTTS-Lada.pt

Then you can inference own texts by the following command:

python3 inference.py -c config_ljs_dap.json -r models/RADTTS-Lada.pt -t test_sentences.txt --vocoder_path models/M_105.pth --vocoder_config_path waveode_config.yaml -o results/ --sampling_method euler --sampling_steps 20