The SVAT project is part of the Connected Health working area and aims to provide a server-based service through which speech and voice pathologies can be automatically analyzed. The algorithms within this modular service are based on technologies of automatic speech recognition, further classifiers, and machine learning approaches.
These technologies are already used at our institute in other application areas and are constantly being further developed. Within the first project phase, they will be integrated, adapted, and implemented in the application area of speech and voice and expanded and tested within the second project phase. The feature algorithms for the automatic analysis of speech and voice are derived from the acoustic properties and characteristics within the disorders and link technology with linguistics and phonetics in the development.
The provision of a service for the automatic analysis of speech and voice disorders opens many areas of application in the research context as well as within the logopedic and speech therapy care chain and contributes to the objectification of the diagnostic process and therapy evaluation.