Yet Another Voice Activity Detection Engine

Voice activity detection (VAD) is a tiny algorithm that monitors a stream of audio for human speech.

Enter Cobra!

Cobra is Picovoice’s VAD engine. Cobra is:

• Cross-platform (Raspberry Pi, NVIDIA Jetson, Android, iOS, Web, Linux, macOS, Windows, …)
• Efficient (Runs on Raspberry Pi Zero with 5% CPU usage)
• Accurate even in harsh noisy environments
• Available on GitHub and free for developers to explore!

Web Demo

Try Cobra’s WebAssembly SDK live here!

Python Demo

Install the demo package:

pip3 install pvcobrademo

Go to Picovoice Console, sign up for a free account, create your AccessKey, and run the demo:

cobra_demo_mic --access_key ${YOUR_ACCESS_KEY}

Benchmark

WebRTC has an excellent VAD that also comes with Python binding. We created an open-source benchmark comparing Cobra to alternatives. The picture below shows a so-called ROC curve. The higher the area under the curve, the better. For example, one can see that Cobra achieves 90+% accuracy (true positive) at a 5-% false alarm rate while WebRTC is just below 50%.

Next Steps

Head to GitHub and start building with Cobra!