Perceived loudness of spatially distributed sound sources

In noise-control engineering, one is often faced with the task of identifying the most problematic of several simultaneous sound sources. Traditionally, this has been done by deriving sound pressure (or intensity) maps by means of a microphone array. This approach does not, however, take psychoacoustic attributes into account. Therefore, a method for deriving loudness maps was developed in an earlier study [Song, Internoise2004, paper 271]. The present experiment investigates to which extent perceived loudness depends on the distribution of individual sound sources. Three loudspeakers were positioned 1.5 m from the centre of the listener’s head, one straight ahead, and two 10 degrees to the right and left, respectively. Six participants matched the loudness of either one, or two simultaneous sounds (narrow-band noises with 1-kHz, and 3.15-kHz centre frequencies) to a 2-kHz, 60-dB SPL narrow-band noise placed in the frontal loudspeaker. The two sounds were either originating from the central speaker, or from the two offset loudspeakers. It turned out that the subjects perceived the noises to be softer when they were distributed in space. In addition, loudness was calculated from the recordings of a microphone (monaural) and a dummy head (binaural) placed at the listening position. The results show that while loudness metrics fared well in predicting perceived loudness for any single-sound condition, they failed to predict loudness for two simultaneous sounds. This suggests that current loudness modelling will have to be extended to take the spatial distribution of sources into account