Abstract:
This article is devoted to the development and analysis of machine learning methods for the automated diagnosis of voice disorders, which is an urgent task in modern medicine and speech therapy. Voice plays a key role in communication, and its pathologies can significantly reduce the quality of life of patients. Traditional diagnostic methods, including visual examination and endoscopy, require the participation of specialists and do not always ensure objectivity. In this regard, the use of machine learning algorithms opens up new opportunities for improving the accuracy of speech disorders detection and diagnosis. The paper examines the main types of voice disorders, including dysphonia, aphonia, phonasthenia, bradylalia, tachylalia, stuttering, dyslalia and rhinolalia. The etiology, symptoms, and existing correction methods are analyzed for each of them. Special attention is paid to acoustic parameters of the voice, such as pitch frequency, jitter, shimmer, and signal-to-noise ratio, which can serve as markers of pathologies.
Keywords:voice disorders, machine learning, acoustic analysis, dysphonia, aphonia, support vector machine, gradient boosting, diagnostics of voice disorders.
Fulltext:
PDF file (375 kB)