Abstract
Detection of Vocal Cord Disorders is one of the essential processes to be determined at an early stage. Nowadays, machine learning methods are mostly used to detect vocal cord disorders. Finding the various parameters associated with existing machine learning is a time-consuming and memory-consuming process. Therefore, a new machine learning method is needed to overcome the current method's drawbacks. Hence, this research work introduces a Sequential Learning Resource Allocation Neural Network (SL-RAN) to overcome existing methods' drawbacks. This work aims to manage voice data precision due to inconsistencies in extracting potential features and dilemmas when dealing with voice signals. Initially, voice signals are preprocessed using the discrete wavelet transform method, and possible features of voice disorder detection are extracted using a Mel Frequency Cepstrum feature extraction method. After removing the features, the proposed Sequential Learning resource allocation neural network (SL-RAN) classifies the types of vocal cord disorders. The SL-RAN process is fine-tuned. Instead of strengthening the voice, the Saarbrucken Voice Database (SVD) database and disorder detection are randomly assigned with Weight. The simulation results show that the proposed model's performance is far better than existing machine learning methods. The sensitivity, specificity, and accuracy of SL-RAN are 96.09%, 94.63%, and 94.35%