# Orpheus TTS Orpheus TTS is an open source (Apache License) text-to-speech LLM created by [[Canopy Labs]]. This model was built on top of Llama-3b. Orpheus supports different languages and different voices. Interestingly, it's possible to add emotions to the text. Those will be conveyed in the generated audio output. ## Local usage To use Orpheus TTS locally, one can use a combination of [[LM Studio]] and applications such as the following ones: - https://github.com/Lex-au/Orpheus-FastAPI - https://github.com/isaiahbjork/orpheus-tts-local I tried the first (Orpheus-FastAPI), which provides a nice UI: ![[Orpheus TTS - orpheus-fastapi ui.png]] I used it within [[Windows Subsystem for Linux (WSL)]] by following these steps: https://github.com/Lex-au/Orpheus-FastAPI?tab=readme-ov-file#fastapi-service-native-installation. I can the model itself locally using [[LM Studio]] for ease of use. ## References - Announcement: https://canopylabs.ai/model-releases - AI models: https://huggingface.co/canopylabs - Source code: https://github.com/canopyai/Orpheus-TTS - Supported emotions: https://github.com/canopyai/Orpheus-TTS/blob/main/emotions.txt