Kardome

Voice commands fail when the room is noisy or several people are talking at once.

Kardome is a voice user interface tool that tackles this by clustering speech signals based on where they come from, so a device can pick up clear commands and deliver clean audio output in challenging environments.

The company sells to manufacturers and OEMs who already have a speech recognition system and want it to be more accurate, rather than to end users directly.

Kardome uses neural network technology to move voice UI away from a cloud-dependent setup.

The result is meant to be secure, real-time and customizable, which matters if you care about latency or about keeping audio off remote servers.

Because it works as an AI speech recognition and noise cancellation layer, Kardome fits into products as a component, not a standalone app.

It is made for hardware and developer teams, so expect an integration project and not a download-and-use experience.