Paper Title
A Criticial Evaluation of Voice Inflictions in Speech Disability With Large Language Model Using Deep Learning Technique
Abstract
An individual's voice might be impacted by diseases of internal organs apart from the vocal folds. This has led to
an increase in the prevalence of vocal issues, which are often disregarded. A good example is the significant interest in
developing an automated system that can differentiate between healthy and disordered voices for use in the early detection
and diagnosis of voice disorders. Consequently, new opportunities in healthcare are opened up by voice analysis helped by
Artificial Intelligence (AI). This research set out to determine which automated speech signal analysis approaches were most
helpful in the diagnosis of voice abnormalities and to analyze numerous modelfor distinguishing between normal and
abnormal speech. Chroma, mel spectrogram, and Mel Frequency Cepstral Coefficient (MFCC) are three aspects of speech
that the traditional modelstakes into account. As a prominent feature extractor in audio processing, MFCCs are utilized
extensively for the purpose of extracting crucial information from voice signals. But MFCC characteristics are often
computed using a single taper window, which has a high variance. This research presents findings from numerous models
that aim to reduce variance in the classification of normal voice and disordered sound using MFCC characteristics. Research
on the early diagnosis and detection of voice problems has focused heavily on the automatic system for classifying sounds as
normal or disordered. Audio recordings of both normal and disordered voices are considered and models for classification of
voice pathology and classification and voice corrections.This research analyzes voice processing models, feature extraction
and selection methods, classification techniques. This research presents a brief survey on voice pathology models based on
deep learning for normal and disordered voice classification and voice correction models. This research is helpful for
numerous academicians and research scholars for designing of efficient models for accurate and efficient voice disorder
classification and voice correction models.
Keywords - Voice Pathology, Voice Processing, Feature Extraction, Feature Selection, Multi Level Classification, Voice
Correction, Deep Learning