You are here

ITS Research Offers New Avenues for Improving Speech Recognition in Wireless Communications

October 18, 2017 by Keith Gremban, Director of the Institute for Telecommunication Sciences

Noise is one of the more vexing communications issues first responders face during emergencies. Emergency scenes are often loud and chaotic, and responders must compete with background noise — sirens, yelling, a roaring fire, severe weather and more — when trying to communicate with the command center and each other.

Improving over-the-air transmissions so that essential verbal communications can be heard above background noise is part of the goal of signal processing, a decades-long specialty of NTIA’s research laboratory, the Institute for Telecommunication Sciences (ITS). This week, ITS is presenting research that offers new avenues for addressing speech separation problems while enabling more efficient use of spectrum. The research could lead to solutions that produce higher speech quality, better speech intelligibility and less distorted sound — while using less bandwidth.

In the world of wireless transmission, signals (audio, video, radio, medical images, radar, GPS, and many others) are processed to achieve certain desired goals, typically a more robust, higher quality, or a more tightly compressed transmission. Signal processing can be applied to any frequency that is being used to transmit a signal. Engineers and researchers involved in this field are focused on finding new and better ways to compress signals for bandwidth efficiency without degrading the perceived quality of the transmitted sound or image. If you could access unlimited bandwidth, quality of transmission would not be a problem. But as more and more transmissions are squeezed into available bandwidth, maintaining the quality of each individual transmission becomes a significant signal processing challenge.

ITS engineer Stephen Voran presented ITS’s most recent advances in signal processing at the 2017 IEEE International Workshop on Applications of Signal Processing to Audio and Acoustics in New York. This biennial workshop brings together top academic, industry, and government researchers to present and discuss their most recent progress, often leading to new and improved signal processing applications being built into all sorts of communications devices.

Beyond the public-safety benefits, the research could lead to higher quality hands-free voice calls in cars and fewer repeat requests in voice controlled systems like Siri, Cortana and Amazon Echo.

This recent advance at ITS exemplifies the successful use of NTIA resources in an applied research program that can open the way for new commercial products while promoting efficiency to allow more shared use of limited spectrum. At the same time, this advance in scientific understanding enables ITS to better support the Federal agencies, state and local Governments, and standards development organizations that look to ITS as a resource for solving their telecommunications concerns.