The aim of the project is to test a model of natural continuous speech that can quantify and distinguish between a) hardware constraints (articulatory, acoustic and perceptual) that have to be implemented in models of automatic speech technology; b) language-specific constraints or software constraints that have to be learn by the system. The model will be tested against a wide number of variables and the correlation between observed and expected values will be assessed. The result of the model will allow to experimentally implement the software and hardware constraints of natural speech on speech synthesis and automatic speech recognition systems to achieve a better modelling of the system and to improve the quality of the results.
|Effective start/end date||8/07/94 → 8/07/97|
- Dirección General de Investigación Científica y Técnica (DGICYT)