huggingface/speech-to-speech: Build voice agents with open-source models

github.com29 min read

GitHub - huggingface/speech-to-speech: Build voice agents with open-source models · GitHub provides an open-source, low-latency voice agent framework that runs entirely locally without requiring API keys or cloud services. The swappable pipeline uses components like Silero VAD, Parakeet TDT, Qwen3.5-4B, and Qwen3-TTS, and supports the OpenAI Realtime protocol over WebSocket and WebRTC. Open article →

More on this topic