The role of fundamental frequency and target segregation in the presence of same‐band and different‐band maskers

Abstract

A number of recent studies on multitalker speech perception have been based on envelope‐modulated, sine‐wave‐excited speech similar to the kind used to simulate the processing used in cochlear implants. Although these modulated sine‐band signals are highly intelligible in isolation, they are typically much harder to segregate in multitalker listening tasks than normal speech. In this experiment, we examined the extent to which the difficulty in segregating these signals could be attributed to the absence of the fundamental frequency F0 that is present in natural speech. Psychometric functions were measured with speech and noise maskers for three types of signals: (1) natural speech; (2) sine‐excited cochlear implant speech (CIS); and (3) F0‐excited CIS, which was generated in the same way as the CIS but with a simulated glottal pulse train as an excitation source. The results show that performance was better with the F0‐excited CIS than with the sine‐excited CIS in all conditions, including those where the target and masking talkers overlapped in frequency. Performance was also generally better when the F0‐excited CIS was generated with F0 values from different‐sex rather than same‐sex talkers. The results confirm the importance of F0 cues in the segregation of competing speech signals.