Automatinė balso sutrikimų vertinimo sistema: sukūrimas ir panaudojimas

Correct diagnosis of the voice disorder is an essential step towards its appropriate treatment and cost control. The aim of this thesis was to create and evaluate an automated voice analysis system, that differentiates normal and pathological voices using sustained vowel and continuous speech and to use it as a potential screening tool for voice disorders. Conclusions: Acoustic voice quality index in Lithuanian language (AVQI-LT) represents significant and marked concurrent validity and high diagnostic accuracy, distinguishing normal and dysphonic voices; No impact of gender and age for AVQI-LT values was revealed concluding that it appears independent of age- and gender-related vocal physiology and anatomy; Comparable results between AVQI-LT and Dysphonia severity index with a high level of validity to discriminate between normal and dysphonic voices were revealed; The AVQI-LT obtained from voice recordings using smartphone microphones revealed almost perfect agreement with studio microphone recordings, thus proving of suitability of smartphone microphone recordings for voice analysis and estimation; Combination of AVQI and Glottal function index measurements significantly improved diagnostic accuracy in differentiating normal vs pathological voice and represented potentially valuable and robust voice screening tool; The elaborated VoiceScreen system represents valid and reliable tool for normal and dysphonic voice discrimination and screening