Abstract

I leverage speech technology to building quantitative models of adult phonetic category perception and infant phonetic learning in humans.  In this talk, I will show how this approach not only provides concrete implementation of cognitive theories and tests of the similarity between machines and humans, but also can reveal blindspots in our current understanding of infant speech processing, leading to the development of new theories.  I’m also hopeful that it will lead to improvements in speech technology, in particular for low/zero-resource technologies.

Biography

I did my PhD in the Laboratoire de Sciences Cognitives et Psycholinguistique (LSCP) at Ecole Normale Superieure (ENS) in Paris under the supervision of Emmanuel Dupoux and Francis Bach.  I’m currently working as a postdoc with Naomi Feldman at the University of Maryland.