zefie/multi-tts repository overview

⁠zefie's Multi-TTS Docker

Currently includes Camb.ai Mars5⁠, Coqui⁠, MyShell.ai OpenVoice⁠, Parler⁠, Suno Bark⁠, and TorToiSe⁠.

⁠Usage:

New 20240829 Fixed an issue where using OpenVoice may overwrite the source speaker wav. Also implemented OpenVoice v2.

20240823 has been removed due to the overwrite bug mentioned above.

New 20240817 The application now runs as the user "app" so if you are upgrading, change your paths from /root/.cache to /home/app/.cache

Example usage (CLI):

docker run --rm -it --gpus all -p 7860:7860 -v multitts-cache:/home/app/.cache -v ./srcwav:/home/app/srcwav zefie/multi-tts:latest

Example usage (Docker Compose):

 zefie-multitts:
    image: zefie/multi-tts:latest
    ports:
      - 7860:7860
    volumes:
      - ./srcwav:/home/app/srcwav
      - multitts-cache:/home/app/.cache
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              device_ids: ['0']
              capabilities: [compute, utility]

  volumes:
    multitts-cache: {} 

For voice cloning, make a folder under "srcwav" with the name you'd like to appear in the list. Then populate that folder with clean wav files of that person speaking. Example (container path): /home/app/srcwav/Tom/tom1.wav

Play around with your source files until you get the voice you desire. The script will import all wavs in said folder. I recommend 16bit Mono, 22050-48000hz for best results.

⁠Building from Source

git clone https://github.com/zefie/multi-tts.git --depth=1 && cd multi-tts && docker build -t multi-tts:latest .

⁠Screenshots