This public repository showcases how to run Rime with LiveKit to build real‑time voice agents. It is designed so customers and developers can quickly understand the purpose, explore the structure, and spin up an example agent that performs a full STT → LLM → TTS loop for back‑and‑forth voice conversations.

• Rime Simple Agent

  • A simple example implementation that demonstrates a complete voice conversation pipeline using LiveKit and Rime. It showcases a seamless integration of Speech-to-Text (STT) → Large Language Model (LLM) → Text-to-Speech (TTS) for natural, bidirectional voice conversations.
  • This is a simple script that serves as a perfect starting point for developers who want to see a basic STT-LLM-TTS implementation and build on top of it.
  • The code is intentionally kept straightforward and well-documented so you can easily understand, download, and run it as a foundation for your own voice agent implementation.

• Rime Multilingual Agent

  • A multilingual voice agent that demonstrates automatic language detection and dynamic voice switching using LiveKit, Deepgram STT, and Rime TTS.
  • The agent detects the user's language in real-time and automatically responds in the same language, supporting English, Spanish, French, and German.
  • This example showcases how to override the STT node to intercept speech events, detect language changes, and dynamically update TTS configuration for natural multilingual conversations.