Perceived Quality Assessment of Speech using EEG Signals

Abstract

Speech is an important component of today’s growing multimedia communication systems. To provide high-quality multimedia communication products or services, quality of perceived speech plays an important role. Better quality of speech ensures that the effort required by the users to correctly perceive the audio is low and subsequently results in low cognitive loading and mental fatigue. Various subjective/objective measures for speech quality assessment such as Mean Opinion Score (MOS), Perceptual Evaluation of Speech Quality (PESQ) etc. have been used. Recently, researchers have reported encouraging results of speech quality assessment based on electroencephalogram (EEG) which is the electrical activity recording the brain signals. It continuously provides the apprehension and sharpness about subject’s listening cognitive states without interruptions caused by directly asking him/her to give the score about the quality of audio.

In this study a series of speech sample is labeled to three different classes based on their quality.