[GitHub Trending] huggingface/speech-to-speech
Scored daily by a customisable AI persona to surface the most relevant engineering leadership news.
Open-source speech-to-speech for local voice agents directly aligns with AI agent trends and is highly actionable.
Hugging Face's speech-to-speech library provides a modular, low-latency voice-agent pipeline (VAD → STT → LLM → TTS) exposed via an OpenAI Realtime-compatible WebSocket API, with every component swappable. It supports local models like Parakeet TDT for STT, Qwen3-TTS for speech output, and can point the LLM slot at hosted providers, Hugging Face Inference Providers, or local vLLM/llama.cpp servers for a fully open stack. The pipeline already powers thousands of Reachy Mini robots in production, and installation is a single pip command with optional backends for Kokoro, Whisper, and ChatTTS.
huggingface