Vous êtes ici : GIPSA-lab > Animation > Séminaires

Séminaire du département Parole et Cognition du 20/07/2017 à 14h00


ASR-based training for eLearning and eHealth

Intervenant : Helmer Strik , Radboud University Nijmegen

Lieu : Salle B 314, Site Ampere , Gipsa Lab 3ème étage, 11 Rue des Maths


Résumé :

From the appearance of the first dictation applications, Automatic Speech Recognition (ASR) technology has been seen as a powerful instrument for processing speech and providing feedback. Over the years, this has generated an impressive amount of research aimed at evaluating ASR technology and its potential for eLearning and eHealth, at improving algorithms to realize more appropriate systems and provide better feedback, and at developing a whole range of research applications and commercial products.

There is a growing demand for computer-based language learning and speech therapy programs that offer speaking practice because this cannot sufficiently be provided in standard classrooms or face-to-face therapy sessions. Given that the quality of 'automatic speech recognition' (ASR) has gradually improved during the last decades, there have been increasing attempts to use ASR for providing speaking practice and feedback in eLearning and eHealth. However, learner speech and pathological speech, often referred to as atypical speech, differ in various ways from (typical) native, non-pathological speech, for instance, with respect to grammar, vocabulary, pronunciation, and dysfluencies. This makes ASR of atypical speech particularly complex and dedicated ASR algorithms often need to be developed for this purpose.

At our lab in Nijmegen, we have been involved in various project on [the use of ASR in] eLearning and eHealth. In this talk, I will present some of the research conducted at our lab in projects on ASR-based eLearning and eHealth (for more info. see, e.g., http://hstrik.ruhosting.nl/projects/ and http://hstrik.ruhosting.nl/publications/).

GIPSA-lab, 11 rue des Mathématiques, Grenoble Campus BP46, F-38402 SAINT MARTIN D'HERES CEDEX - 33 (0)4 76 82 71 31