Revolutionary Dysarthria Analysis Through Convolutional Neural Networks

Dysarthria occurs when there are impairments in the muscles necessary for speech, including those in the lips, tongue, voice box, and breathing muscles. This condition can result in unclear speech, even in individuals who are skilled communicators. Dysarthria can range from mild speech difficulties, where speech is hard to understand, to severe cases, where speech is completely unclear. A method for distinguishing between different types of speech involves using recordings and advanced computational techniques such as convolutional neural networks (CNNs). The strengths of CNNs are leveraged to construct a highly effective model for detecting dysarthria in speech samples. The dataset utilized includes recordings from individuals with and without dysarthria. The results indicate that the CNN model performs exceptionally well, achieving precision, recall, and F1-scores of 0.99 for both categories. With an accuracy of 0.99, the model demonstrates a high level of proficiency in differentiating between speech affected by dysarthria and normal speech. These findings highlight the effectiveness and utility of this CNN-based approach in diagnosing and assessing dysarthria. It offers valuable potential for assisting healthcare professionals and enhancing personalized speech therapy interventions.