The ecological utility of extended high frequencies for speech recognition

Recent studies have demonstrated that extended high frequencies (EHFs; >8 kHz) in speech are audible and useful for speech recognition. In this paper I will review recent and ongoing work in our lab examining audibility of EHF cues in speech and the conditions under which these cues are or are not useful. Mismatches in head orientations between a target talker (facing the listener) and masker talkers (not facing the listener) influence the utility of EHFs. For example, EHFs provide a speech recognition benefit when masker talkers are facing away from the listener, but not when maskers are facing the listener. Spatial separation of target and maskers may also influence this relationship. Speech tokens with higher EHF levels are more likely to be intelligible. Our data indicate that EHF cues in speech are useful in real-world auditory scenes, suggesting the loss of EHF hearing, which typically begins in early adulthood, could contribute to speech-in-noise difficulties.