Development of a model to predict the likelihood of complaints due to assorted tone-in-noise combinations

This paper develops a model to predict if listeners would be likely to complain due to annoyance when exposed to a certain noise signal with a prominent tone, such as those commonly produced by heating, ventilation, and air-conditioning systems. Twenty participants completed digit span tasks while exposed in a controlled lab to noise signals with differing levels of tones, ranging from 125 to 1000 Hz, and overall loudness. After completing the digit span tasks under each noise signal, from which task accuracy and speed of completion were captured, subjects were asked to rate level of annoyance and indicate the likelihood of complaining about the noise. Results show that greater tonality in noise has statistically significant effects on task performance by increasing the time it takes for participants to complete the digit span task; no statistically significant effects were found on task accuracy. A logistic regression model was developed to relate the subjective annoyance responses to two noise metrics, the stationary Loudness and Tonal Audibility, selected for the model due to high correlations with annoyance responses. The percentage of complaints model showed better performance and reliability over the percentage of highly annoyed or annoyed.