Objective speech quality assessment with non-intrusive method for narrowband speech
Jing Wang, Juan Luo, Shenghui Zhao · 2008
A non-intrusive objective assessment method is proposed to estimate the quality of output speech without the input reference speech based on narrowband speech test database. From clean speech Perceptual Linear Predictive (PLP) features are extracted and clustered by Gaussian Mixture Model (GMM) as an artificial reference model. Input speech is separated into three classes, for which the consistency measures between features of the test speech signal and the GMM reference model are calculated and mapped to an objective speech quality score using Support Vector Regression (SVR) method. Experiment results show that the proposed method has a higher objective to subjective correlation degree than ITU-T P.563 within 6 narrowband MOS-labeled test databases.