Orpheus TTS FastAPI Windows One Click Installer
A new and exciting Text-to-Speech tool called Orpheus TTS has been recently released. Orpheus TTS is an open-source system built on the Llama-3b backbone, showcasing the potential of using LLMs for speech synthesis.
Key Features of Orpheus TTS:
Human-Like Speech: Offers natural intonation, emotion, and rhythm superior to many closed-source models.
Zero-Shot Voice Cloning: Clone voices without prior fine-tuning.
Guided Emotion and Intonation: Control speech characteristics with simple tags.
Low Latency: Approximately 200ms streaming latency, reducible to 100ms with input streaming.
The community has developed a sleek FastAPI WebUI with an OpenAI API compatible endpoint, making it easier to use the model. You can find this repo and installation steps here: https://github.com/Lex-au/Orpheus-FastAPI.
I've created a forked version that integrates LM Studio out of the box and extends the context window to 8192. This version uses the GGUF model, which is less resource-intensive than the full FP16 model and should run on <4GB VRAM. You can find the forked repo here: https://github.com/TheLocalLab/Orpheus-FastAPI-LMStudio.
For Patreon or YouTube members, I've prepared a one-click Windows installer and step-by-step instructions to make setup as easy as possible.
You can join as a member or purchase the installer individually here: https://www.patreon.com/posts/orpheus-tts-w-lm-124899399.
Feel free to join our Discord and share your thoughts on the new model, or if you have any improvement ideas for the forked repo!
Buy On Patreon
While I improve the store, you can purchase these items or sign up for a membership on Patreon - https://www.patreon.com/TheLocalLab.