Effects of entropy in real-world noise on speech perception in listeners with normal hearing and hearing loss

Hearing aids show more benefit in traditional laboratory speech-in-noise tests than in real-world noisy environments. Real-world noise comprises a large range of acoustic properties that vary randomly and rapidly between and within environments, making quantifying real-world noise and using it in experiments and clinical tests challenging. One approach is to use acoustic features and statistics to quantify acoustic properties of real-world noise and control for them or measure their relationship to listening performance. In this study, the complexity of real-world noise from different environments was quantified using entropy in both the time- and frequency-domains. A distribution of noise segments from low to high entropy were extracted. Using a trial-by-trial design, listeners with normal hearing and hearing loss (in aided and unaided conditions) repeated back sentences embedded in these noise segments. Entropy significantly affected speech perception, with a larger effect of entropy in the time-domain than the frequency-domain, a larger effect for listeners with normal hearing than for listeners with hearing loss, and a larger effect for listeners with hearing loss in the aided than unaided condition. Speech perception also differed between most environment types. Combining entropy with the environment type improved predictions of speech perception above the environment type alone.

Automated summary

This study investigated the effect of real-world noise on speech perception and found that entropy significantly influenced speech perception, with a larger effect in the time-domain compared to the frequency-domain. Additionally, listeners with normal hearing were more affected by entropy than listeners with hearing loss, and the effect was larger for hearing loss listeners in the aided condition. Speech perception also varied across different environment types. Combining entropy with environment type improved predictions of speech perception.