Vector Quantization and MFCC based Classification of Dysfluencies in Stuttered Speech

Abstract

Stuttering also known as stammering is a speech disorder that involves disruptions or dysfluencies in speech. The observable signs of dysfluencies include repetitions of syllable or word, prolongations, interjections, silent pauses, broken words, incomplete phrases and revisions. The repetitions, prolongations and interjections are important parameter in assessing the stuttered speech. The objective of the paper is to classify the above mentioned three types of dysfluencies using Mel-Frequency Cepstral Coefficients (MFCC) and Vector Quantization (VQ) framework. For each dysfluency MFCC features are extracted and quantized to a number of centroids using the K-means algorithm. These centroids represent the codebook of dysfluencies. The dysfluencies are classified according to the minimum quantization distance between the centroids of each dysfluency and the MFCC features of testing sample.