Modulation masking in a speech recognition task for normal-hearing and hearing-impaired subjects

Abstract

The effect of masking in the modulation domain was determined in a consonant recognition task. It is known from psychoacoustical experiments [T. Houtgast, J. Acoust. Soc. Am. 85, 1676–1680 (1989)] that modulation detection can be masked by another modulation. An experiment was designed that tries to verify whether detection of the modulations in a speech signal is essential for speech recognition. The goal of this experiment was to determine whether stochastic modulations can mask the modulations in a speech signal and disturb speech recognition as effectively as filtering out speech modulations [R. Drullman et al., J. Acoust. Soc. Am. 95, 1053–1064 (1994)]. The masker used was a stochastic narrow-band signal with minimal inherent fluctuations. The masker signal was, in every critical band, multiplied with the speech modulations. Several masking conditions were created with different bandwidths and center frequencies. The center frequencies were all lying in the region of modulation frequencies which are important for speech (4–20 Hz). Normal-hearing subjects and subjects with sensorineural hearing loss participated in the experiment. The confusion matrices were analyzed by means of multidimensional scaling techniques. The results of the two groups of subjects will be compared with the results obtained with a psychoacoustic model.