Vliv nelinearní odezvy v kochlee na vjem řeči v šumu

Abstract

This thesis investigates the role of cochlear compression in speech perception, particularly in noisy environments, using an advanced auditory model. The primary focus is on the analysis of speech in multitalker babble noise by using an auditory model composed with a dual resonance non-linear (DRNL) filterbank. A key aspect of this study is examining how the overall signal intensity, meaning the intensity of speech and babble noise with constant signal-to-noise ratio (-5 dB), influences these predictions. The research employs a systematic approach, comparing the auditory model's predictions and showing the similarity between the model outputs to speech and speech plus babble noise with actual experimental listening test results from the literature. The currently used listening test results reveal a notable improvement in speech perception in noise at intensities up to about 70 dB sound pressure level (SPL). In contrast, the model predictions presented in this thesis do not fully show improvement with increasing sound levels. This discrepancy highlights the complexity of cochlear compression and suggests that additional factors beyond cochlear compression may contribute to speech perception in noisy environments. The findings are pivotal in understanding the extent to which cochlear compression contributes to this improvement.

This thesis investigates the role of cochlear compression in speech perception, particularly in noisy environments, using an advanced auditory model. The primary focus is on the analysis of speech in multitalker babble noise by using an auditory model composed with a dual resonance non-linear (DRNL) filterbank. A key aspect of this study is examining how the overall signal intensity, meaning the intensity of speech and babble noise with constant signal-to-noise ratio (-5 dB), influences these predictions. The research employs a systematic approach, comparing the auditory model's predictions and showing the similarity between the model outputs to speech and speech plus babble noise with actual experimental listening test results from the literature. The currently used listening test results reveal a notable improvement in speech perception in noise at intensities up to about 70 dB sound pressure level (SPL). In contrast, the model predictions presented in this thesis do not fully show improvement with increasing sound levels. This discrepancy highlights the complexity of cochlear compression and suggests that additional factors beyond cochlear compression may contribute to speech perception in noisy environments. The findings are pivotal in understanding the extent to which cochlear compression contributes to this improvement.