AI4ESP · Artificial intelligence for early speech production

Lines connect the coordinator with its partners.

Project objective

Babbling is a critical milestone of infant speech development. In only 12 months, infants explore how to move their mouth, tongue, and vocal cords to produce increasingly complex sounds before eventually producing their first words. Investigating early vocal development in both typical and atypical infant populations requires meticulous efforts involving the time-consuming and labor-intensive process of manually transcribing children’s linguistic productions. The present action, at the crossroads between speech processing, machine learning, and language acquisition, proposes to 1) build an open-source automatic universal phone recognizer for infant vocalizations to accelerate language acquisition research; 2) design an innovative measure to assess when early vocalizations reflect the infant’s ambient language(s); and 3) apply the developed methodology to the study of typical and atypical infant populations. By combining large-scale multilingual corpora and state-of-the-art automatic speech processing tools, our first objective is to endow researchers with a novel tool to advance our understanding of infants’ early vocal development. Our second objective is to shed new light on the simple yet controversial question: When do infants start exhibiting distinct babbling patterns across different languages? Beyond its theoretical implications, our proposal holds clinical significance as we will apply the developed tools and measures to investigate early vocal markers associated with speech disorders, constituting our third objective.

Original text from CORDIS.

Participants

• MAX-PLANCK-GESELLSCHAFT ZUR FORDERUNG DER WISSENSCHAFTEN EV · MUNCHENCoordinatorGermany

Links

• View on CORDIS
• DOI: 10.3030/101199931