HVAC and Refrigeration (HVAC&R) equipment designers would benefit if there were better noise criteria for optimization of noise control treatments. Development of a model that can predict the average annoyance ratings of residential HVAC&R equipment sound is described. Three sets of subjective tests were performed, and the results from part of the third test are described. Three rating exercises were designed for the 60 subjects who participated. In each of the three parts the 50 sounds were a mixture of recordings and recordings that were digitally modified to decrease the correlation between sound metrics. In Part A the sounds tended to be louder, in Part B quieter sounds were rated and in Part C the sounds covered a wider range of loudness, similar to that used in the previous two tests. Thirteen sound quality metrics were tested as parameters in a linear model that predicts average annoyance ratings. The best performing model contained a tone-adjusted level, a spectral balance, a tonality and a roughness metric. The model was also used to predict average annoyance ratings in previously conducted tests and R2 values ranged from 0.90 to 0.94. Ongoing work includes examination of model deficiencies to identify potential sound attribute levels that cause actual ratings to be higher than those predicted.
Sound quality, Signal modification, HVAC noise
Acoustics and Noise Control
Date of this Version