Effects of age, modulation rate, and modulation depth on sentence recognition in speech-modulated noise

Abstract

Speech is often heard in amplitude modulated backgrounds when speech is glimpsed during momentary masker dips. Increased masker modulation depth provides more improvement in the signal-to-noise ratio of glimpsed speech, but also improves noise modulation detection for potentially greater modulation masking. Mechanisms of modulation masking and glimpsing may also depend on the overlap of the modulation spectra of the speech and noise. This study employed modulation filtering and noise amplitude compression to investigate the combined effects of noise modulation depth and modulation rate on speech recognition. Younger normal-hearing (YNH), older normal-hearing (ONH), and older hearing-impaired (OHI) adults listened to sentences in noise that were spectrally shaped to control for individual hearing thresholds. A second YNH group listened to sentences with the same spectral shaping as OHI listeners. Sentence recognition generally improved with greater noise modulation depth, especially at higher modulation rates. Results suggest that speech recognition for all groups is maximized when speech modulations <8 Hz are preserved, when noise modulation occurs at rates higher than this range, and with greater noise modulation depth. OHI listeners benefit similarly from these conditions, but their poorer overall performance may be due to reduced sensation levels. [Work supported by NIH/NIDCD.]