Skip to content

Latest commit

 

History

History
73 lines (48 loc) · 1.13 KB

README.md

File metadata and controls

73 lines (48 loc) · 1.13 KB
Raw

LLM-TTS-Speak

Chat with LLM that speak on your audio, supports full offline mode with LibreTranslate and Ollama.

Quickstart example:

npm install
docker compose up --profile piper up -d
npm run start

For Piper, get the voices from piper initial release page on github, get the .onnx and .onnx.json files from the zip and put it inside the models folder, then put the voice name in the .env file, for example:

PIPER_VOICE_NAME=en-us-ryan-high

Docker

VOICEVOX CPU

Start

docker compose --profile voicevox-cpu up -d

Stop

docker compose --profile voicevox-cpu down

VOICEVOX GPU

Start

docker compose --profile voicevox-gpu up -d

Stop

docker compose --profile voicevox-gpu down

Piper

Start

docker compose --profile piper up -d

Stop

docker compose --profile piper down

TODO

  • Write a better Docs
  • LLM uses memory/memory management
  • Make the code better