Search Results

This paper describes the development of a semantic differential test paradigm for evaluating product sound quality by a population of human subjects. The design of this specific experiment was also structured to test for a variety of expected common experimental errors. An investigation on the dependence of presentation order was executed by implementing both a fully balanced and a partially balanced design, and differences in subject populations were tested by independently testing two geographically separate populations. In order to conduct this testing, a general semantic-differentials graphical user interface (GUI) was written in MATLAB to present the specific stimuli to each subject in a unique specific sequence and extract a variety of semantic differential ratings for later statistical analyses. A discussion of the advantages and disadvantages for Semantic Differentials approach is presented as well as a discussion of the confidence in the resulting subject dataset.

This paper describes the development of a model for the predicted quality assessment of power seat mechanisms. This model was based on the statistical correlation of objective popular sound quality metrics (and their associated distribution statistics) to a set of subjective results obtained in a previously reported [1] semantic differentials evaluation experiment. More important then the model itself, the greatest surprises were in the statistical analyses of the subjective response data and the statistical significance and independence of the various commercial software generated sound quality metrics. Step-wise regression was used to identify a sub-set of independent variables that best fit the subjective response data and determine the appropriate model coefficients. This resultant model showed significantly better correlation to this population of subjective response data than other overall sound quality metrics e.g., Loudness, UBA, Pleasantness, etc.