Cobra Voice Activity Detection

Detect the presence of human voice activity in audio streams

Accurate, lightweight, and production-ready VAD

Start Building for FreeWork with an expert
Probability of Voice

Highly accurate — proven scientifically.

Check out the open-source benchmark that indicates Cobra VAD outperforms Google’s well-known WebRTC VAD with wide margins. Cobra VAD achieves a much higher true-positive rate at any false-positive operating point.

2023-01-25T15:28:36.467670image/svg+xmlMatplotlib v3.5.1,


Cobra VAD runs even on microcontrollers or even offline within your browser. After starting the demo, turn off your internet. It keeps working, right?

Start Building
Probability of Voice
Start Building

Build with Picovoice SDKs Instantly

Build voice-activated experiences with the SDK of your choice. Cobra VAD supports Python, Android, iOS, JavaScript, Rust, and C.

o = pvcobra.create(access_key)
while True:
is_voiced =
Build with Python
Cobra o = new Cobra(accessKey);
while(true) {
float isVoiced =
Build with Android
let o =
Cobra(accessKey: accessKey)
while true {
let isVoiced =
Build with iOS
let o =
await CobraWorker.create(
(isVoiced) => {
// callback
const processor =
await processor.start()
Build with JavaScript
let o = Cobra::new(access_key)
loop {
let is_voiced = o
Build with Rust
while (true) {
Build with C

Deploy Anywhere

Add voice activity detection to your existing platforms. Expand later without worrying about the support. Cobra VAD runs on web browsers, mobile platforms, single board computers, microcontrollers, on-premise, or even in the cloud.

Start Building

Learn more about Cobra Voice Activity Detection Engine

  • What is Voice Activity Detection?

    Voice activity detection (VAD) is a technology used to detect the presence of human speech within an audio signal containing speech and noise. That is why it is also known as speech activity detection, speech detection, or voice detection. VAD is essential to enable automatic speech recognition (ASR). We initially developed Cobra as an internal tool and then made it publicly available since no computationally efficient and accurate VAD was available.

  • How does Cobra Voice Activity Detection perform better than WebRTC?

    Typical VAD methods use learned statistical models such as the Gaussian mixture model, just like the most popular WebRTC VAD. WebRTC VAD is good, computationally efficient and works for streaming audio signals. On the other hand, Cobra VAD uses deep learning. Cobra’s proprietary algorithm is developed by applying Picovoice’s expertise in on-device voice recognition. Therefore, Cobra VAD could achieve higher accuracy and run across platforms.

  • Does Cobra Voice Activity Detection carry any security flaws and leak data?

    Cobra VAD processes voice data on-device, resulting in private, HIPAA and GDPR compliant experiences. The data could be processed within a web browser, on a mobile application, in an IoT device, a laptop or a server. On your terms!

  • Which platforms does Cobra Voice Activity Detection support?

    1. Single Board Computers: Raspberry Pi, NVIDIA Jetson, and BeagleBone
    2. Mobile Devices: Android and iOS
    3. Web Browsers: Chrome, Safari, Firefox, and Edge
    4. Desktop and Servers: Linux, macOS, and Windows
  • What can I build with Cobra Voice Activity Detection?

    Cobra VAD can be used alone or paired with other engines, Rhino Speech-to-Intent and Leopard Speech-to-Text for several use cases such as call-centre automation, telemarketing, audio conferencing, filtering, voice assistants and so on. Don’t forget to check out Voice Command and Control, Search by Voice and Speech Analytics use cases to learn more about various voice AI applications with Cobra VAD.

  • How do I get technical support for Cobra Voice Activity Detection?

    Picovoice docs, blog, Medium posts, and GitHub are great resources to learn about voice recognition, Picovoice engines, and how to start building voice-activated products. Picovoice also offers GitHub community support to all Free Tier users.

  • How can I get informed about the updates and upgrades?

    Version changes appear in the Picovoice Newsletter, LinkedIn, and Twitter. Subscribing to GitHub is the best way to get notified of the patch releases. If you enjoy building with Cobra, don’t forget to give it a star when you’re on GitHub!