Second ISCA/DEGA Tutorial and Research Workshop on Perceptual Quality of Systems

Berlin, Germany
September 4-6, 2006

Non-Intrusive SOM-Based Speech Quality Assessment for Telephony Applications

Abdulhussain E. Mahdi

Department of Electronic and Computer Engineering, University of Limerick, Limerick, Ireland

A non-intrusive method for speech quality assessment in telephony applications is proposed and its performance evaluated. The method involves measuring perception-based objective auditory distances between the voiced parts of the processed (degraded) speech signal to appropriately matching references extracted from a pre-formulated codebook. The codebook is formed by optimally clustering large number of parametric speech vectors extracted from a database of clean speech records, using an efficient data-mining tool known as the Self-Organizing Map (SOM). The auditory distances are mapped into an objective Mean Opinion listening quality scores (MOS_LQO). Three domain transformation techniques have been utilized to provide perception-based, speakerindependent parametric representation of the speech: a Perceptual Linear Prediction (PLP) model, a Bark Spectrum (BS) analysis and Mel-Frequency Cepstrum Coefficients (MFCC). Reported evaluation results show that the proposed method provides high correlation with subjective listening quality scores (MOS_LQS), yielding accuracy similar to that of the ITU-T P.563 while maintaining a relatively low computational complexity. Results also demonstrate that the method outperforms the PESQ in assessing the quality of speech degraded by channel impairments.

Full Paper

Bibliographic reference.  Mahdi, Abdulhussain E. (2006): "Non-intrusive SOM-based speech quality assessment for telephony applications", In PQS-2006, 123-130.