Preferences for reproduced sound can be related to acoustical measurements made at a listener's ears, but to understand fully what contributes to a listener's preference choices, it is necessary also to determine what the most salie salient auditory attributes are that might predict nt those choices. This problem is important in sound quality evaluation, since physical measures of a single stimulus typically do not deal with the contextual effects that can strongly influence the results of p preference tests, such as differences between musical programs and their reference interaction with other experimental variables such as recording technique technique. A model is . Presented for predicting preferences for multichannel sound programs based not only upon acoustic acoustical measurements, but also upon psychological data, which includes both direct ratings al of selected auditory attributes and pairwise stimulus dissimilarity judgments. The comprehensive view of preference prediction afforded by this approach enables cross va validation between results of complementary listening tests, and can potentially account for lidation contextual effects based upon interaction between multichannel sound recording systems and the musical program material used to evaluate the quality of those systems systems.
展开▼