- [2025/11] We supported Python inference with ONNX model on Linux, macOS thanks to Guy Nicholson!
- [2025/11] We supported Golang on Linux, macOS and Windows with usage of the prebuilt-libs thanks to hylarucoder!
- [2025/11] We supported Java on Linux, macOS, Windows, Android with usage of the prebuilt-libs thanks to ZhangYang(arthasking123)!
- [2025/07] 🎉 Exciting news! TEN VAD is now integrated into k2-fsa/sherpa-onnx, thanks to the fantastic work by Fangjun Kuang! You can now achieve more precise speech segment extraction and enjoy an enhanced ASR experience! Refer to the documentation and give it a try!
- [2025/07] We supported Python inference on macOS and Windows with usage of the prebuilt-libs!
- [2025/06] We finally released and open-sourced the ONNX model and the corresponding preprocessing code! Now you can deploy TEN VAD on any platform and any hardware architecture!
- [2025/06] We are excited to announce the release of WASM+JS for Web WASM Support.
- Welcome to TEN
- TEN Hugging Face Space
- Introduction
- Key Features
- Developers Testimonial
- Installation
- Quick Start
- TEN Ecosystem
- Ask Questions
- Citations
- License
TEN is an open-source framework for conversational voice AI agents.
TEN Ecosystem includes TEN Framework, Agent Examples, VAD, Turn Detection and Portal.
| Community Channel | Purpose |
|---|---|
 |
Follow TEN Framework on X for updates and announcements |
| Follow TEN Framework on LinkedIn for updates and announcements | |
 |
Join our Discord community to connect with developers |
 |
Join our Hugging Face community to explore our spaces and models |
 |
Join our WeChat group for Chinese community discussions |
Important
Star TEN Repositories ⭐️
Get instant notifications for new releases and updates. Your support helps us grow and improve TEN!
huggingface-demo.mp4
You are more than welcome to Visit TEN Hugging Face Space to try VAD and Turn Detection together.
TEN VAD is a real-time voice activity detection system designed for enterprise use, providing accurate frame-level speech activity detection. It shows superior precision compared to both WebRTC VAD and Silero VAD, which are commonly used in the industry. Additionally, TEN VAD offers lower computational complexity and reduced memory usage compared to Silero VAD. Meanwhile, the architecture's temporal efficiency enables rapid voice activity detection, significantly reducing end-to-end response and turn detection latency in conversational AI systems.
The precision-recall curves comparing the performance of WebRTC VAD (pitch-based), Silero VAD, and TEN VAD are shown below. The evaluation is conducted on the precisely manually annotated testset. The audio files are from librispeech, gigaspeech, DNS Challenge etc. As demonstrated, TEN VAD achieves the best performance. Additionally, cross-validation experiments conducted on large internal real-world datasets demonstrate the reproducibility of these findings. The testset with annotated labels is released in directory "testset" of this repository.
Note that the default threshold of 0.5 is used to generate binary speech indicators (0 for non-speech signal, 1 for speech signal). This threshold needs to be tuned according to your domain-specific task.
Developers can reproduce the performance comparison PR curves for TEN VAD and Silero VAD on the open-source testset (as shown in the figure above) by executing the following script on Linux x64 with a simply one line of code. The output figure will be saved in the same directory as the script.
cd ./examples
python plot_pr_curves.py
As illustrated in the figure below, TEN VAD rapidly detects speech-to-non-speech transitions, whereas Silero VAD suffers from a delay of several hundred milliseconds, resulting in increased end-to-end latency in human-agent interaction systems. In addition, as demonstrated in the 6.5s-7.0s audio segment, Silero VAD fails to identify short silent durations between adjacent speech segments.
We evaluated the RTF (Real-Time Factor) across five distinct platforms, each equipped with varying CPUs. TEN VAD demonstrates much lower computational complexity and smaller library size than Silero VAD.
| Platform | CPU | RTF | Lib Size | ||
|---|---|---|---|---|---|
| TEN VAD | Silero VAD | TEN VAD | Silero VAD | ||
| Linux | AMD Ryzen 9 5900X 12-Core | 0.0150 | / | 306KB | 2.16MB(JIT) / 2.22MB(ONNX) |
| Intel(R) Xeon(R) Platinum 8253 | 0.0136 | ||||
| Intel(R) Xeon(R) Gold 6348 CPU @ 2.60GHz | 0.0086 | 0.0127 | |||
| Windows | Intel i7-10710U | 0.0150 | / | 464KB(x86) / 508KB(x64) | |
| macOS | M1 | 0.0160 | 731KB | ||
| Web | macOS(M1) | 0.010 | 277KB | ||
| Android | Galaxy J6+ (32bit, 425) | 0.0570 | 373KB(v7a) / 532KB(v8a) | ||
| Oppo A3s (450) | 0.0490 | ||||
| iOS | iPhone6 (A8) | 0.0210 | 320KB | ||
| iPhone8 (A11) | 0.0050 | ||||
TEN VAD provides cross-platform C compatibility across five operating systems (Linux x64, Windows, macOS, Android, iOS), with Python bindings optimized for Linux x64, with wasm for Web.
TEN VAD operates on 16kHz audio input with configurable hop sizes (optimized frame configurations: 160/256 samples=10/16ms). Other sampling rates must be resampled to 16kHz.
"We selected TEN VAD because it provides faster and more accurate sentence-end detection in Japanese compared to other VADs, while still being lightweight and fast enough for live use." - LiveCap,Hakase shojo.
"TEN VAD's overall performance is better than Silero VAD. Its high accuracy and low resource consumption helped us improve efficiency and significantly reduce costs." - Rustpbx.
git clone https://github.com/TEN-framework/ten-vad.git
The project supports five major platforms with dynamic library linking.
| Platform | Dynamic Lib | Supported Arch | Interface Language | Header | Comment |
|---|---|---|---|---|---|
| Linux | libten_vad.so | x64 | Python, C, Java, Go | ten_vad.h ten_vad.py ten_vad.js TenVad.java |
|
| Windows | ten_vad.dll | x64, x86 | C, Java, Go | ||
| macOS | ten_vad.framework | arm64, x86_64 | C, Java, Go | ||
| Web | ten_vad.wasm | / | JS | ||
| Android | libten_vad.so | arm64-v8a, armeabi-v7a | C, Java | ||
| iOS | ten_vad.framework | arm64 | C | 1. not simulator 2. not iPad |
-
numpy (Version 1.17.4/1.26.4 verified)
-
scipy (Version >= 1.5.0)
-
scikit-learn (Version 1.2.2/1.5.0 verified, for plotting PR curves)
-
matplotlib (Version 3.1.3/3.10.0 verified, for plotting PR curves)
-
torchaudio (Version 2.2.2 verified, for plotting PR curves)
-
Python version 3.8.19/3.10.14 verified
Note: You could use other versions of above packages, but we didn't test other versions.
The lib only depend on numpy, you have to install the dependency via requirements.txt:
pip install -r requirements.txt
For running demo or plotting PR curves, you have to install the dependencies:
pip install -r ./examples/requirements.txt
Note that if you did not install libc++1 (Linux), you have to run the code below to install it:
sudo apt update
sudo apt install libc++1
Note: For usage in python, you can either use it by git clone or pip.
- Clone the repository
git clone https://github.com/TEN-framework/ten-vad.git
- Enter examples directory
cd ./examples
- Test
python test.py s0724-s0730.wav out.txt
- Install via pip
pip install -U --force-reinstall -v git+https://github.com/TEN-framework/ten-vad.git
- Write your own use cases and import the class, the attributes of class TenVAD you can refer to ten_vad.py
from ten_vad import TenVad
You have to download the onnxruntime packages from the microsoft official onnxruntime github website. Note that the version of onnxruntime must be higher than or equal to 1.17.1 (e.g. onnxruntime-linux-x64-1.17.1.tgz).
You can check the official ONNX Runtime releases from this website. And for example, to download version 1.17.1 (Linux x64), use this link. After extracting the compressed file, you'll find two important directories:include/ - header files, lib/ - library files
1) cd examples_onnx/python
2) ./build-and-deploy-linux.sh --ort-path /absolute/path/to/your/onnxruntime/root/dir # For Linux. If macOS, run build-and-deploy-macos.sh instead.
Note 1: If executing the onnx demo from a different directory than the one used when running build-and-deploy-linux.sh, ensure to create a symbolic link to src/onnx_model/ to prevent ONNX model file loading failures.
Note 2: The ONNX model locates in src/onnx_model directory.
Note 3: For macOS, run build-and-deploy-macos.sh instead.
- Node.js (macOS v14.18.2, Linux v16.20.2 verified)
- Terminal
1) cd ./examples
2) node test_node.js s0724-s0730.wav out.txt
TEN VAD provides comprehensive Java support with JNI (Java Native Interface) bindings for all major platforms.
- Java 8 or higher
- Native libraries in
lib/directory - JNI headers
# Compile Java source, note if JNA library is not installed, you should download it first. For example, you can download JNA library and include it here. You can also export it to the CLASSPATH environment variables
wget https://repo1.maven.org/maven2/net/java/dev/jna/jna/5.13.0/jna-5.13.0.jar # I just picked a random version
javac -encoding UTF-8 -cp jna-5.13.0.jar -d . include/TenVad.java examples/TestTenVad.java
# Run example in the project root directory
java -cp .:jna-5.13.0.jar TestTenVad examples/s0724-s0730.wav examples/out.txt
# Run example in the examples directory
java -cp ..:../jna-5.13.0.jar TestTenVad s0724-s0730.wav out.txtimport com.ten.vad.TenVad;
public class VADExample {
public static void main(String[] args) {
// Create VAD instance
TenVad vad = new TenVad(256, 0.5f);
// Process audio frame
short[] audioFrame = new short[256]; // 16ms at 16kHz
// ... fill audioFrame with audio data ...
TenVad.VadResult result = vad.process(audioFrame);
System.out.println("Probability: " + result.getProbability());
System.out.println("Voice detected: " + result.isVoiceDetected());
// Clean up
vad.destroy();
}
}- Linux: Requires
libc++1package - Windows: Ensure Visual C++ Redistributable is installed
- macOS: No additional requirements
- Android: Use Android NDK for native library integration
public class TenVad {
// Constructor
public TenVad(int hopSize, float threshold)
// Process audio frame
public VadResult process(short[] audioData)
// Get library version
public static String getVersion()
// Cleanup
public void destroy()
}
public static class VadResult {
public float getProbability() // [0.0, 1.0]
public int getFlag() // 0 or 1
public boolean isVoiceDetected() // true if voice detected
}TEN VAD provides Golang support for Linux, macOS and Windows.
cd examples/go-tenvad
go build -o tenvad .
./tenvad
cd examples/go-tenvad
export LD_LIBRARY_PATH=../../lib/Linux/x64:$LD_LIBRARY_PATH # For Windows and macOS, export the corresponding lib path instead
go run .
Located in examples/ directory or examples_onnx/ (for ONNX usage on Linux):
- Linux: build-and-deploy-linux.sh
- Windows: build-and-deploy-windows.bat
- macOS: build-and-deploy-mac.sh
- Android: build-and-deploy-android.sh
- iOS: build-and-deploy-ios.sh
Runtime library path configuration:
- Linux/Android: LD_LIBRARY_PATH
- macOS: DYLD_FRAMEWORK_PATH
- Windows: DLL in executable directory or system PATH
- Modify platform-specific build scripts
- Adjust CMakeLists.txt
- Configure toolchain and architecture settings
- Navigate to examples/ or examples_onnx/ (for ONNX usage on Linux)
- Execute platform-specific build script
- Configure dynamic library path
- Run demo with sample audio s0724-s0730.wav
- Processed results saved to out.txt
The detailed usage methods of each platform are as follows
- Clang (e.g. 6.0.0-1ubuntu2 verified)
- CMake
- Terminal
Note that if you did not install libc++1, you have to run the code below to install it:
sudo apt update
sudo apt install libc++1
1) cd ./examples
2) ./build-and-deploy-linux.sh
You have to download the onnxruntime packages from the microsoft official onnxruntime github website. Note that the version of onnxruntime must be higher than or equal to 1.17.1 (e.g. onnxruntime-linux-x64-1.17.1.tgz).
You can check the official ONNX Runtime releases from this website. And for example, to download version 1.17.1 (Linux x64), use this link. After extracting the compressed file, you'll find two important directories:include/ - header files, lib/ - library files
1) cd examples_onnx
2) ./build-and-deploy-linux.sh --ort-path /absolute/path/to/your/onnxruntime/root/dir
Note 1: If executing the onnx demo from a different directory than the one used when running build-and-deploy-linux.sh, ensure to create a symbolic link to src/onnx_model/ to prevent ONNX model file loading failures.
Note 2: The ONNX model locates in src/onnx_model directory.
Note 3: For ONNX example builds see examples_onnx/README.md.
- Visual Studio (2017, 2019, 2022 verified)
- CMake (3.26.0-rc6 verified)
- Terminal (MINGW64 or powershell)
1) cd ./examples
2) Configure "build-and-deploy-windows.bat" with your preferred:
- Architecture (default: x64)
- Visual Studio version (default: 2019)
3) ./build-and-deploy-windows.bat
- Xcode (15.2 verified)
- CMake (3.19.2 verified)
1) cd ./examples
2) Configure "build-and-deploy-mac.sh" with your target architecture:
- Default: arm64 (Apple Silicon)
- Alternative: x86_64 (Intel)
3) ./build-and-deploy-mac.sh
- NDK (r25b, macOS verified)
- CMake (3.19.2, macOS verified)
- adb (1.0.41, macOS verified)
1) cd ./examples
2) export ANDROID_NDK=/path/to/android-ndk # Replace it with your NDK installation path
3) Configure "build-and-deploy-android.sh" with your build settings:
- Architecture: arm64-v8a (default) or armeabi-v7a
- Toolchain: aarch64-linux-android-clang (default) or custom NDK toolchain
4) ./build-and-deploy-android.sh
Xcode (15.2, macOS verified) CMake (3.19.2, macOS verified)
- Enter examples directory
cd ./examples
- Creates Xcode project files for iOS build
./build-and-deploy-ios.sh
-
Follow the steps below to build and test on iOS device:
3.1. Use Xcode to open .xcodeproj files: a) cd ./build-ios, b) open ./ten_vad_demo.xcodeproj
3.2. In Xcode IDE, select ten_vad_demo target (should check: Edit Scheme → Run → Release), then select your iOS Device (not simulator).
3.3. Drag ten_vad/lib/iOS/ten_vad.framework to "Frameworks, Libraries, and Embedded Content"
-
(in TARGETS → ten_vad_demo → ten_vad_demo → General, should set Embed to "Embed & Sign").
-
or add it directly in this way: "Frameworks, Libraries, and Embedded Content" → "+" → Add Other... → Add Files →...
-
Note: If this step is not completed, you may encounter the following runtime error: "dyld: Library not loaded: @rpath/ten_vad.framework/ten_vad".
3.4. Configure iOS device Signature
-
in TARGETS → ten_vad_demo → Signing & Capabilities → Signing
-
Modify Bundle Identifier: modify "com.yourcompany" to yours;
-
Specify Provisioning Profile
-
-
In TARGETS → ten_vad_demo → Build Settings → Signing → Code Signing Identity:
-
Specify your Certification
3.5. Build in Xcode and run demo on your device.
-
-
| Project | Preview |
|---|---|
| ️TEN Framework Open-source framework for conversational AI Agents.  |
 |
| TEN VAD Low-latency, lightweight and high-performance streaming voice activity detector (VAD).  |
|
| ️ TEN Turn Detection TEN Turn Detection enables full-duplex dialogue communication.  |
 |
| TEN Agent Examples Usecases powered by TEN. |
 |
| TEN Portal The official site of the TEN Framework with documentation and a blog.  |
 |
TEN VAD is available on these AI-powered Q&A platforms. They can help you find answers quickly and accurately in multiple languages, covering everything from basic setup to advanced implementation details.
| Service | Link |
|---|---|
| DeepWiki |  |
| ReadmeX |  |
@misc{TEN VAD,
author = {TEN Team},
title = {TEN VAD: A Low-Latency, Lightweight and High-Performance Streaming Voice Activity Detector (VAD)},
year = {2025},
publisher = {GitHub},
journal = {GitHub repository},
howpublished = {https://github.com/TEN-framework/ten-vad.git},
email = {[email protected]}
}
This project is licensed under Apache 2.0 with certain conditions. Refer to the "LICENSE" file in the root directory for detailed information. Note that pitch_est.cc contains modified code derived from LPCNet, which is BSD-2-Clause and BSD-3-Clause licensed, refer to the NOTICES file in the root directory for detailed information.