PUB - Publikationen an der Universität Bielefeld

Robots in social interaction need to be able to communicate with humans in a human-like manner by understanding and using their language. One important first element of speech and language understanding is the ability to parse spoken utterances into words. But this ability is not innate and needs to be developed by infants within the first years of their life. So far almost all computational speech processing systems neglected this bootstrapping process. Here we propose a model for early infant word learning embedded into a layered architecture comprising phone, phonotactics and syllable learning. Our model uses raw acoustic speech as input and aims to learn the structure of speech unsupervised on different levels of granularity. In addition to previous work we describe how our model can be used to learn a basic syntax model in interaction. To close the loop between speech structure acquisition and production we further outline how our system can be combined with a speech imitation system.