acm-header
Sign In

Communications of the ACM

ACM TechNews

Research Team Develops Voice Localization Techniques for Smart Speakers


View as: Print Mobile App Share:
Reverse-triangulation of vocal sounds.

Researchers at the University of Illinois at Urbana-Champaign's Coordinated Science Laboratory (CSL) have created a system with which the Amazon Alexa smart speaker can localize a user's whereabouts based on their voice.

Credit: University of Illinois at Urbana-Champaign Coordinated Science Lab

Researchers at the University of Illinois at Urbana-Champaign's Coordinated Science Laboratory (CSL) have created a system that uses the Amazon Alexa smart speaker's microphone array in combination with echoes of one’s voice to localize a user's whereabouts.

CSL's Sheng Shen said the VocLoc system uses an "align-and-cancel algorithm” that identifies the directions from which it receives each voice signal and uses them to reverse-triangulate the location of the person speaking.

The system also spontaneously learns some aspects of the room's geometry, to help with that triangulation.

CSL's Romit Roy Choudhury said, "Location can help Alexa in improving speech recognition, since different speech vocabularies and models can be loaded."

From University of Illinois at Urbana-Champaign Coordinated Science Lab
View Full Article

 

Abstracts Copyright © 2020 SmithBucklin, Washington, DC, USA


 

No entries found

Sign In for Full Access
» Forgot Password? » Create an ACM Web Account