Can formant shifts and effort cues enhance boundary tone perception in whispered speech?

Abstract

Whispered speech holds cues to speech melody, in spite of the absence of F0. Shifts in the locations of formant peaks have been forwarded as a main cue. Whispering speakers, however, may convey high versus low boundary tones signalling questions versus statements without shifting their formants. Would the addition of formant shifts enhance these natural productions and improve question/statement classification in whisper? Moreover, multiple acoustic correlates tend to vary with pitch or intonation conditions in whispered speech, and may function as listener cues. Here, an attempt was made to better understand the function of one of these ‘secondary’ cues: intensity. Results show that formant shifts may improve performance, but not dramatically, and that intensity seems more useful when coding increased effort than when being higher across the board to compensate for reduced audibility in whisper