32 Auditory Scene Analysis : A Prerequisite for Loudness Perception

Auditory scene analysis (ASA) refers to the ability of the auditory system to organize perceptually a sound mixture into distinct auditory streams, each stream ideally corresponding to a single sound source (see Bregman 1990 for a review). Level cues can be used to promote perceptual segregation of auditory objects. As such, temporal sequences of pure tone bursts with the same frequencies but with alternating levels can be perceived as segregated as soon as a level difference is introduced between the tones (van Noorden 1975). The inverse relation, i.e., the dependency of loudness on grouping processes, has been much less described in the literature, and the extent to which ASA processes can influence loudness perception remains largely undetermined. Jeng (1992) measured the relative loudness of two distinct sounds (speech and a simulated jack-hammer sound). She found that the total loudness did not follow the Zwicker power-spectrum loudness model (Zwicker and Fastl 1990). This is similar to the situation of a masking noise on speech. As shown by Fletcher (Fletcher and Munson 1933, 1937; Fletcher 1938), speech can be masked by noise, and as the noise level is increased, the loudness of the speech is decreased. From the spectral point of view, the total loudness should be found by summing up all the specific loudness components. In fact, this is not the case, because of grouping. McAdams et al. (1998) provided an argument supporting the assumption that the loudness of an auditory event can be influenced by the perceptual organization. The stimuli used in their experiments were alternating sequences of two identical bursts with no silent gap between them and played at different levels, well above threshold. According to Warren et al. (1972) and van Noorden (1975), these sequences lead to the perception of a continuous sound upon which is superimposed an additional intermittent stream of bursts. This phenomenon suggests that the