This is a SG2300x-adapted demo of EmotiVoice(@NetEase Youdao) and OpenVoice(@MyShell.ai).
git clone https://github.com/ZillaRU/EmotiVoice-TPU.git
bash download.sh
bash prepare.sh
Run the following script.
python3 demo_page.py
Three tabs for TTS, TTS+tone color cloning and tone color cloning are shown.
- Install the dependencies:
pip3 install fastapi pydub uvicorn[standard] pyrubberband
. - Then, run
uvicorn openai_api:app --reload --host 0.0.0.0 --port [port_number]
, the service will be available athostip:port_number
in few seconds. The way to call the service is exactly the same as using OpenAI's TTS service.
For more detailed usage, please refer to the README from the original repo of EmotiVoice and OpenVoice.
The torch version higher than 2.3 might lead to an ERROR due to poor compatibility for ARM platform.