Logo image
Back
Effects of entropy in real-world noise on speech perception
Abstract

Effects of entropy in real-world noise on speech perception

Erik Jorgensen, Yu-hsiang Wu and Inyong Choi
The Journal of the Acoustical Society of America, Vol.150(4), pp.A303-A303
10/2021
DOI: 10.1121/10.0008361

View Online

Abstract

Traditional tests of speech-in-noise perception use artificial background noise, leading to results that do not generalize to real-world scenarios. There is a significant need for improved ecological validity in laboratory and clinic speech-in-noise tests, particularly with respect to the types of noise used. However, the acoustic complexity and random nature of real-world noise poses challenges to its use and the interpretation of results. In this study, the complexity of real-world noise from 8 environments was systematically quantified in the time and frequency domains using entropy. Then, segments of noise distributed in entropy from low to high were extracted and combined with target sentences at fixed signal-to-noise ratios. Ten normal-hearing subjects were tested on 200 sentences-in-noise across the entropy distribution in both the time and frequency domains in a trial-by-trial design. Linear mixed effect models showed a significant effect of entropy in the time and frequency domains on number of words correct per trial. There was a systematic improvement in speech perception with increases in entropy variance in both time and frequency domains. Entropy in real-world noise affects speech perception even when the overall signal-to-noise ratio is fixed. Experiments using real-world noise should consider this effect in their design and interpretation.

Details

Metrics

70 Record Views
Logo image