← Discover
Discover · GitHub · Speech to Speech↑ +227 today

Open source native voice agents

The speech-to-speech library developed by Hugging Face allows creating local voice agents using open source models. This Python-based tool allows developers to build real-time voice interaction systems that run on the device.

Updates

  • September 7, 2026: Stars 12,310 → 13,072, latest release v1.0.0 (September 6, 2026).
  • August 12, 2026: Stars 11,283 → 12,310, latest release v0.2.12 (August 5, 2026).
  • August 6, 2026: Stars 10,774 → 11,283, latest release v0.2.12 (August 5, 2026).
  • August 4, 2026: Stars 10,402 → 10,774, latest release v0.2.11 (August 3, 2026).

What you get

  • Low latency modular audio line
  • OpenAI Realtime compatible WebSocket support
  • Opportunity to work locally on different hardware

Installation

Basic setup
pip install speech-to-speech
Installation from source code
git clone https://github.com/huggingface/speech-to-speech.git
cd speech-to-speech
uv sync

Running it

Starting the server
pip install speech-to-speech
export OPENAI_API_KEY=...
speech-to-speech
Connecting with the client
python scripts/listen_and_play_realtime.py --host 127.0.0.1 --port 8765

If you don't write code

🤖 Paste this into your AI agent (Claude Code · Codex · Antigravity)

I want to setup my own local voice agent using this tool. What are the basic steps I need to follow to create a low latency audio pipeline using VAD, STT, LLM and TTS components? With what command can I stand up the server and connect with an OpenAI Realtime compatible client?

Related dictionary terms

Who it is forIt is for developers who want to develop native and customizable voice interaction systems on their own hardware.
LicenseApache-2.0

Links

TreScout did not build this tool · we found it in GitHub trends and wrote it up. This page describes the repository as of 2026-07-29: The star count and our text belong to that day, the repository may have changed since. Check the repository link for the current state. This page was machine-translated from the Turkish original · the Turkish version prevails.