HEARDEVCOMP
Language learning in humans is based on the implementation of a biological process of extraction and selective processing of acoustic parameters contained in speech signals, in particular amplitude and frequency modulations. Although previous studies show that the efficiency of auditory processing of these cues—present from birth—improves until adulthood, the stages of this development remain poorly defined. In this project, we suggest a new interdisciplinary combination of hearing measurement methods coupled with computational models to access the parameters of this developmental timeline from infancy to adulthood. Each individual tested will undergo a battery of auditory tests to characterize physiological auditory encoding and speech perception abilities. The characterization of the auditory encoding will associate behavioral (psychophysical, adapted in babies) and neurophysiological (EEG) responses of synthetic signals reproducing the time-frequency modulations of speech. The speech perception capacities will be evaluated by subjective tests of intelligibility measurement of real speech signals in difficult natural listening conditions (e.g. spatialized listening 'cocktail party').
The primary objective of this project is to establish a proof-of-concept for the use of computational models of the auditory system to exploit behavioral and neurophysiological data collected in infants and adults in order to infer latent auditory coding parameters for these two populations, in relation to speech perception abilities observed in everyday life. Ultimately, this work aims to develop new biomarkers of typical and atypical speech and language development, allowing a better understanding of the impact of childhood and adult deafness on auditory development and speech perception in everyday life.
IRCAM's Team : Sound Perception and Design
