Prof Sarah Hawkins
S2S Co-ordinator
Dept of Linguistics, University of Cambridge, UK
Research Interests: Acoustic phonetics, FPD, perception, adverse conditions, formant synthesis
Email: sh110@cam.ac.uk
Website: http://www.ling.cam.ac.uk/people/sarah/
I work mainly in acoustic phonetics, speech perception, and speech synthesis. However, I have wide interests in speech production and perception, reflecting my belief that neither can be studied in isolation from the other.
My early work (in roughly chronological order) concerned:
- the development of temporal coordination and rhythm in children’s speech
- acoustic correlates of distinctive features, especially nasal vowels and consonants
- voice quality and speech movement.
My current focus is on how acoustic-phonetic fine detail varies systematically to reflect the entire linguistic (grammatical and pragmatic) structure of an utterance, and the contribution of this systematically-varying, informative fine phonetic detail to how we understand speech. In October 2003 I began a three-year Leverhulme Trust Major Research Fellowship to extend this work on speech perception into neuropsychology and computational modelling. See my publications list for some conference presentations, which are currently being written up.
One outcome of the Leverhulme-funded research is Sound to Sense (S2S), a Marie Curie Research Training Network involving 13 institutions in 10 countries. We aim to bring together research on fine phonetic detail with computer modelling of human and machine speech recognition. This a training programme for pre-and post-doctoral students. It starts 1st May 2007 and runs until 2011.
An application of my speech perception work is in exploring the potential of systematic variation in fine phonetic detail for increasing the intelligibility of synthetic speech, especially in adverse listening conditions.
I have contributed to the development of a commercial text-to-speech system for British English (the Infovox system) and am a member of the ProSynth synthesis group.
Involved in: » Project 3: Perceptual salience of FPD 1 - Segmentation of speech into words and phrases » Project 10: Hybrid episodic-abstract computational modelling: HSP
