Audio Demo
Hush
Real-time background speaker suppression for Voice AI.
Use headphones for best results.
Before — Raw (Noisy Environment)
Original audio with background noise and competing speaker
After — Denoised (Hush Output)
Enhanced audio with background speech suppressed
8 MB model · Runs fully on CPU in real time ·
<1 ms per 10 ms frame · 10,000+ hours of training data
← Back to repository