Quantitative algorithms for the calculation of loudness or A-weighted sound power levels are useful to product companies for meeting regulatory or advertising needs. Most consumer product companies recognize however that these measures are inadequate for determining customer satisfaction with product sound. Newer algorithms that compute quantities such as roughness and sharpness have been introduced and are now fairly widely available. A common inadequacy of all such algorithms is their inability to incorporate cognitive aspects of product sound. This presentation will describe a procedure for converting consumer judgments of product sounds into design requirements for the components. The procedure involves jury testing, using combinations or ``mixes'' of component sounds. The experimental design, jury training, experimental protocol, and response analysis are described. Finally, an example of the way the results of the procedure are converted to design goals will be presented.