Prosody in whispered speech: retrieving alternative cues to pitch
06 / 2010 - 05 / 2014
The goal of this study is to systematically investigate how intonation (speech melody) is conveyed in whispered speech. A speaker who whispers fails to produce the most important cue to intonation perception, i.e. the fundamental frequency. Assuming that the speaker attempts to accommodate the listener during communication, it is likely that the whispering speaker will compensate for the loss of that cue by exploiting (and exaggerating) alternative acoustic cues to pitch. Both the presence and nature of alternative cues to intonation in whispered speech, and their relevance for listeners, will be investigated. This study will focus on two frequent and cross-linguistically relevant functions of intonation: marking of information structure, and clause typing. The intended outcome of this study is a model of intonation in whispered (Dutch) speech for these two functions. The results may contribute to a better understanding of the fundamental relation between mental representations of language and their realizations in speech. The results may furthermore be used for improved synthesis of whispered speech, with applications in text-to-speech systems, and for new ways of coding pitch information in cochlear implant processors.
Als we spreken, bepaalt de snelheid waarmee onze stembanden trillen de zinsmelodie. Door die melodie kunnen we vragen stellen, woorden accentueren of emotie tonen. Bij fluisteren trillen de stembanden niet. Kunnen we al fluisterend tóch melodie gebruiken? En hoe dan?