TEN Framework
Open-source real-time voice-agent framework — home of TEN VAD, a fast, lightweight voice activity detector.
About TEN Framework
TEN Framework is an open-source framework for building real-time, multimodal voice-agent conversational AI. It emphasizes low-latency, customizable experiences with RTC and WebSocket support, and is home to TEN VAD — a high-performance, lightweight voice activity detector — alongside turn-detection tooling. Backed by Agora and the TEN Community.
Features
- Open-source framework for real-time voice-agent conversational AI
- TEN VAD — low-latency, high-performance, lightweight voice activity detection
- Turn detection for natural conversational turn-taking
- Real-time, low-latency multimodal conversations over RTC and WebSocket
- Documentation, blog, and an active GitHub repository
What is TEN VAD?
TEN VAD is the framework's open-source Voice Activity Detector: a fast, lightweight model that detects when a user is speaking, enabling responsive, real-time voice agents. It ships with TEN Framework and is available on GitHub.
Who Is It For?
- Developers and teams building real-time, multimodal voice agents
- Engineers who need an open-source VAD (TEN VAD) and turn detection
Use Cases
- Real-time multimodal voice-agent experiences
- Voice activity detection and turn-taking with TEN VAD
- RTC- and WebSocket-based conversational interactions
Get started with TEN Framework and TEN VAD on GitHub.
Open source
TEN VAD is an open-source Voice Activity Detector designed for low-latency, high-performance, and lightweight applications. It is suitable for developers and researchers looking to integrate efficient voice detection capabilities into their projects, enhancing audio processing workflows.