Spontaneous Speech Database for the Romanian Language with Medical Applicability

The research in the field of spontaneous speech for Romanian language, with its direct applicability in the medical process of patient remote monitoring and in helping the dyslexics and the dyspraxics, is an unexplored domain, with high potential for worthy results in speech recognition area. The goal is to achieve performance in helping persons with disabilities and common patients. This paper describes the statistical results and the achievements obtained in the field of spontaneous speech recognition, beginning with the new Romanian corpus, built from scratch with words and triphones. The definition and use of conventions is an important aspect in order to reach high standards of performance with applicability to further research. The challenge is to create a spontaneous speech recognition tool for Romanian language with medical applicability for the benefit of persons with difficulties in mobility, communication and even common people.