Restore disturbed speech to improve intelligibility, comprehensibility and communication – OLINPIC
What if tomorrow you were deprived of your ability to communicate ? What would you like as a solution? Currently, writing or a computer can replace your speech or voice, but perhaps you would dream of having your communication capacity returned to you? This is the aim of the OLINPIC project : to restore communication by automatically modifying disturbed speech units thanks to combine expertise from humanities, computer science and health sciences.
Communication disability may concern people with speech production disorders or second language learners that experience articulation imprecisions that affect speech intelligibility. Lack of intelligibility necessarily impacts the quality of life by causing a communication handicap and consequently social isolation and low self-esteem. Speech communication is an essential element of the social bond and involves extremely complex speech production and perception mechanisms. In this project, we consider that speech variablity plays a crucial role in these mechanisms allowing communication across a wide range of acoustic conditions and speech characteristics. In its adaptative function, variability ensure comprehensibility whatever the condition of speech production, transmission and perception. Nevertheless, despite speakers' compensation strategies and listeners' restoration skills, sometimes communication fails as a result of lack of intelligibility and/or comprehensibility. Speech variability, both at the segmental and prosodic level is a result of many factors including linguistic ones and those related to the speaker’s profile (pathological vs. l2 learner vs. control). Given this situation, the major issue of our project is to pinpoint those units that are variable due to pathological or L2 speech profiles i.e. distinguish them from units that are linguistically variable due to prosodic or other linguistic constraints (e.g. co-articulation; speaking styles...) for optimal, linguistically motivated restoration of speech
Project coordination
Muriel Lalain (Université Aix-Marseille)
The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.
Partnership
LNPL Université Toulouse - Jean Jaurès
IRIT Université Toulouse 3 - Paul Sabatier
CHUT-DRI Direction de la Recherche et Innovation
LPL Université Aix-Marseille
LIA Laboratoire Informatique d'Avignon
Help of the ANR 652,941 euros
Beginning and duration of the scientific project:
December 2024
- 48 Months