Open Voice

Open Voice

Open Voice is an open-source voice middleware that lets you have natural conversations with any Large Language Model. Instead of typing, you speak. Instead of reading, you listen.

Key Features

- Local-First: Speech recognition, voice synthesis, and LLM inference run entirely on-device. No API keys required. No cloud dependencies. Your conversations stay private.

- Model Agnostic: Works with any LLM provider—Ollama, llama.cpp, OpenAI, Anthropic, or your own custom endpoints.

- Smart Routing: Automatically routes queries to the right model based on complexity. Simple questions go to fast models; complex tasks go to capable ones.

- Interruption Handling: Speak anytime—Open Voice listens, stops the current response, and processes your new input naturally.

Why Open Voice?

Voice AI today is either locked in proprietary platforms, requires complex infrastructure, or limits you to specific models. Open Voice is different—a simple, local-first middleware that gives you full control without vendor lock-in.

Talk to any LLM.