Simple method for predicting the distraction due to speech and babble

Abstract

Task-irrelevant speech is a general problem in various work environments. The speech scenarios vary a lot due to different numbers of simultaneous speakers, overall sound level, and signal-to-noise ratio. There is no objective method to assess the distraction of such sound environments using measurements. Our purpose was to develop a method (a simple equation) to objectively predict the distraction due to sound environments containing speech or babble using variables available in standard sound level meters. We conducted three psychoacoustic experiments. Experiment 1 (N = 10) was a pilot study. Experiment 2 (N = 41) was the main study, where the method was developed. Experiment 3 (N = 43) was conducted to validate the method of Experiment 2. We used stimuli that consisted of both multi-voice speech (0 − 12 simultaneous speakers) and steady-state broadband masking noise. The signal-to-noise ratio of speech was varied (-5 to +25 dB) as well as the overall level of stimuli (40, 50, or 60 dB L_Aeq). Participants were asked to imagine how distracted they would be if they had to do private work. All three experiments provided similar evidence. We developed a method that predicted the subjective distraction, based on measured values of L_Aeq and L_A5 - L_A95 with very high correlation (r_P=0.96). Experiment 3 showed that the method also predicts the distraction of foreign language speech. The method can be applied in the objective assessment of distraction in work environments, since the variables are available in standard sound level meters. Further validation of the method is needed both in laboratory and in workplaces.