U.S. flag

An official website of the United States government

Official websites use .gov
A .gov website belongs to an official government organization in the United States.

Secure .gov websites use HTTPS
A lock ( ) or https:// means you’ve safely connected to the .gov website. Share sensitive information only on official, secure websites.

Communications Technology Laboratory

Public Safety Communications Research Division

Speech Analytics for Public Safety

Share

Back to the PSCR 2020 Portal
Speech Analytics for Public Safety_On-Demand Session
Speech Analytics for Public Safety_On-Demand Session
As emerging first responder communicative assistive technologies become available, the need for testing and improving speech analytic technologies for first responder scenarios becomes increasingly more important. First responder acoustic scenarios can include noisy background sounds and changes in speech. In 2019, researchers, through a NIST speech analytics evaluation project called OpenSAT, applied their speech analytic technologies to audio that contains public safety response background noise that were subjected to participants playing an interactive fire rescue game. The intent was to induce increased vocal effort and occasional expression of urgency by game players in response to the noise. The game players were subjected to background sounds from actual first responder events. Time constraints were also injected to create a sense of urgency. The changes in speech combined with the background sounds create communicative conditions that can be challenging for speech analytic technologies. The first responder simulation through game-playing could represent real communicative conditions for first responders at critical times during an actual emergency.

Correction to Content: *To clarify, it is possible for Automatic Speech Recognition systems to have excessive insertions, which are similar to false positives in keyword search, but fall outside the 0-100% range to over 100% word error rate.

Click here to view this presentation as a PDF. 

Speaker

  • Fred Byers, Information Technology Laboratory

Back to the PSCR 2020 Portal

Created May 12, 2020, Updated November 1, 2020