In the design of Recreational Vehicle generators, a particular challenge arises from marketing and engineering teams' desire to ensure that their products meet “best in class” sound quality characteristics. Furthermore, it is desirable to know these characteristics in measurable engineering terms in the product design stage, preferably before prototypes are built and tested. Using a combination of product engineering knowledge and sound manipulation techniques, this paper shows how several generator sounds were produced. These new hybrid real/ synthetic sounds along with actual competitive generator sounds were then used for consumer jury tests to determine their preferences. The sounds include a target “best in class” sound based on mechanical design criteria. Since published A-weighted sound levels can be a useful marketing tool, in one study a few different sounds were presented to juries at varying levels of loudness. However, other specific sound quality characteristics were addressed using sounds adjusted to levels of equal loudness so that paired comparison juries would not focus on the loudness differential among sounds.Ranking as well as merit values were obtained from the paired comparison data collected from generator owners. An RV generator sound quality lexicon was generated using critical listening to name the attributes. Attributes were found valid since consumers had little difficulty recognizing some discrete characteristics in these sounds. Sensory to instrumental metric correlation was obtained by measuring the specific loudness values for the attributes in each sound and then rank-ordering the attributes in isolation. A tight relationship was found to exist between the metrics and the rankings of the lexical entries. Using these “attribute metrics” a model for consumer liking was produced which yielded a very high coefficient of correlation.