Correlating Sound Quality Metrics and Jury Ratings

An investigation was conducted into how a large group of sound quality metrics might be used to predict user reactions to sounds from a particular type of product, as expressed in terms of product-specific attributes such as ratings of “acceptability” and perceived “quality” of the product. We assume that a jury study has already been conducted for such attributes, producing rating values for various product sounds, and that the objective is to determine which metrics or combinations of metrics can best be used to predict user judgments for the sounds of different versions of the product. The basic methodology employs the use of principal components analysis to group the large number of sound quality metrics into just a few orthogonal (principal) components or factors composed of a weighted sum of the original metrics. A “metrics profile” is then computed for each sound based on the resulting principal components, followed by the creation of a transformation matrix to convert between mean jury ratings and the metrics profile. The expected performance of this transformation in predicting jury ratings from the metrics profile is then assessed. The procedure is illustrated using an example drawn from yard maintenance equipment.