Objective evaluation of car noise by means of support vector machine

D. Q. Sha, F. L. Jiao, K. Liu, F. S. Gu, A. D. Ball, J. Jiang

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Sound quality analysis basically involves using tedious subjective evaluations, which involve with the problems related to the reliability and repeatability of the evaluation data as well as the time and cost. What is needed is an objective method for predicting subjective evaluation with good accuracy. In this paper, the support vector machine (SVM) has been employed to build models for the two descriptors: annoyance and muffle of the loudness-adjusted car interior noise. The nonmetric multidimensional scaling (NMDS) and principal component analysis (PCA) have been used as data pre-processor. By comparing the results with artificial neural network (ANN) and multivariate linear regression (MLR), it is confirmed that the SVM is an effective method for the modeling in the objective evaluation.

Original languageEnglish
Title of host publicationInstitute of Noise Control Engineering of the USA - 35th International Congress and Exposition on Noise Control Engineering, INTER-NOISE 2006
Pages4779-4788
Number of pages10
Volume7
Publication statusPublished - 1 Dec 2006
Externally publishedYes
Event35th International Congress and Exposition on Noise Control Engineering - Honolulu, United States
Duration: 3 Dec 20066 Dec 2006
Conference number: 35

Conference

Conference35th International Congress and Exposition on Noise Control Engineering
Abbreviated titleINTER-NOISE 2006
Country/TerritoryUnited States
CityHonolulu
Period3/12/066/12/06

Fingerprint

Dive into the research topics of 'Objective evaluation of car noise by means of support vector machine'. Together they form a unique fingerprint.

Cite this