Heat pump noise: Determination and modelling of preference-equivalent levels

Air-to-water heat pumps are increasingly replacing traditional heating systems because of their environmental and energy benefits. However, noise emitted by their fans and compressors can cause complaints, even when regulatory limits are met. In this study, two listening experiments were conducted to examine heat pump noise by quantifying it as level penalties for equivalent loudness and preference. The study investigates how the preference-equivalent level differs from the loudness-equivalent level compared to a reference stimulus, and which psychoacoustic parameters best model the resulting level adjustments. Recordings from two heat pumps were auralised into a suburban residential setting and adjusted to an initial level of 60 dB(A). A two-alternative forced-choice (2-AFC) method with a 1-up-1-down rule is used to determine the point of subjective equality (PSE) for loudness and preference. Statistical analyses show significant differences between loudness-equivalent and preference-equivalent levels for eight of twelve stimuli. The median values further indicate that achieving equal preference relative to the reference generally requires lower sound pressure levels than achieving equal loudness, with the magnitude of this level offset varying across stimuli. Combining the results with psychoacoustic parameter analyses, two models based on linear regression were developed. This linear approach was chosen to ensure straightforward interpretability and robustness for the given stimulus set. Perceived loudness can be modelled by the psychoacoustic parameter loudness, while preference is best modelled by a linear combination of sharpness and roughness. These findings emphasise the need for perceptual metrics beyond A-weighted sound pressure levels when evaluating and optimising heat pump noise.