VoIP Quality Prediction Model by Bio-Inspired Methods
Book chapter abstract
Authors/Editors
Strategic Research Themes
No matching items found.
Publication Details
Author list: Triyason T., Valaisathien S., Vanijja V., Kanthamanon P., Chan J.H.
Publisher: Hindawi
Publication year: 2015
Start page: 95
End page: 116
Number of pages: 22
ISBN: 9780128017432; 9780128015384
ISSN: 0146-9428
eISSN: 1745-4557
Languages: English-Great Britain (EN-GB)
Abstract
Voice over Internet Protocol (VoIP) quality measurement modeling is an active field of research. An objective and nonintrusive approach is preferred because it is faster and easier than a subjective or intrusive method. This chapter provides a brief overview of the bio-inspired methods used thus far in VoIP speech quality modeling. Then it compares several models such as Perceptual Evaluation of Speech Quality (PESQ)/E-model, nonlinear surface regression, neural network, and REPTree to model the conversational quality. Simulated data sets are generated by varying network impairments (packet loss and delay), codecs, languages, and gender to build and test the models. The bio-inspired neural network and the decision-tree-based REPTree models show highly reliable results for both network and nonnetwork impairment cases. In addition, the accuracy of the model is not only dependent on codec and network impairment, but the nonnetwork factors also have an impact on the performance of a speech quality prediction model. ฉ 2015 Elsevier Inc. All rights reserved.
Keywords
Bio-inspired, Nonintrusive
Contact Information