About
Built in India, for Indian calls.
packethush builds voice isolation for Indian phone calls: a library that runs on your own servers and keeps only the caller's voice, before speech recognition hears it.
A caller rings from a busy street. Someone beside them is talking too. The call reaches you as 8 kHz phone audio, and your speech engine writes down both voices. With packethush, it hears only the caller.
What we hold to.
- 01
Measured, not claimed.
Every result on this site says how it was measured, and the evaluation kit runs the same test on your own calls.
How we measure → - 02
What you test is what you run.
The C library, its Python, Go and Rust packages and the FreeSWITCH module return the same audio, sample for sample.
- 03
Your calls stay yours.
packethush runs inside your servers. Your callers' audio is cleaned there and never comes to us.
What stays in your network → - 04
Made for Indian calls.
8 kHz phone lines, a second voice in the room, the street outside: the calls we build for and test on.
Hear it →
Talk to us.
Tell us about your calls, languages and volumes. We reply by email to set up a call.
Request access