A Comprehensive Intelligent System for Classifying Dysarthric-speech Using Ai Technics

Speech is fundamental to human communication, Millions worldwide face challenges due to speech disorders, such as dysarthria—a neurological condition impairing muscle control required for speech. Dysarthric speech, characterized by slurred, slow, and less intelligible patterns, often results from conditions like stroke, and cerebral palsy profoundly impacting individuals’ quality of life. Current diagnostic practices rely heavily on subjective evaluations by speech-language pathologists, which, pose challenges such as inconsistency, resource-intensiveness, and variability in results. This paper presents an intelligent system designed to classify speech as dysarthric or non-dysarthric, addressing these limitations by leveraging advanced AI and machine learning techniques. The system incorporates a comprehensive workflow, beginning with audio data preprocessing to ensure consistency and integrity, followed by feature extraction to represent speech signals numerically. Key features include Mel Frequency Cepstral Coefficients (MFCCs), spectral contrast, and chroma features, chosen for their ability to capture spectral, tonal, and textural characteristics of speech. These features were input into machine learning models, including support vector machines (SVM), neural networks (NN), and convolutional neural networks (CNN). The CNN model demonstrated remarkable performance, achieving an impressive 98 percent accuracy when using MFCCs, outperforming the NN (97 percent) and SVM (80 percent) models.The system’s generalizability was validated using a diverse dataset, the TORGO corpus, which includes recordings of individuals with varying degrees of dysarthria. The incorporation of segmentation, preprocessing, and model training ensured the system’s adaptability to real-world variability in speech patterns. This paper not only demonstrates the technical feasibility of automating dysarthria detection but also emphasizes its transformative potential in clinical settings. By reducing the diagnostic burden on clinicians and offering scalable, real-time solutions, this AI-driven approach can improve access to speech disorder evaluation and treatment. This pioneering work lays the foundation for practical clinical applications, offering a reliable, and scalable tool to enhance the diagnosis and treatment of dysarthria. By fostering improved communication and social integration, this research contributes meaningfully to the lives of individuals with speech disorders while advancing the field of speech pathology.