A binary masking technique for isolating energetic masking in speech perception

Abstract

When a target speech signal is obscured by interfering speech, two distinct types of masking contribute to the resulting degradation in the intelligibility of the target talker: energetic masking caused by overlap in the time-frequency distribution of energy in the two voices, and informational masking caused by the listener’s inability to correctly segregate the acoustic elements of the two voices into distinct streams. This study attempted to isolate the effects of energetic masking on multitalker speech perception with ideal time-frequency binary masks that retained those spectro-temporal regions of the acoustic mixture that were dominated by the target speech but eliminated those regions that were dominated by the interfering speech. This procedure removed the same phonetic information from the target speech that would ordinarily be lost due to energetic masking, but eliminated the possibility for the kinds of target-masker confusions that are thought to produce informational masking. The results suggest that energetic masking may play a surprisingly small role in the overall masking that occurs in certain types of multitalker speech signals. They also indicate that the number of competing talkers has a much greater influence than target-masker similarity on the amount of energetic masking that occurs in a multitalker stimulus.