An Effect of the Coherence between Envelopes across Frequency Regions on the Perception of Roughness

Auditory roughness arises when rapid beats can be perceived. It is considered to be linked to the amount of amplitude fluctuation within a given auditory filter, around a “roughest” modulation frequency of 70 Hz. The way roughness builds up if sounds have energy spread over a large frequency region is unclear. When two sinusoidally amplitude-modulated (SAM) tones with different carrier frequencies but a same modulation frequency are added, conditions with “co-phasic” envelopes are rougher than those with “anti-phasic” envelopes. This could be accounted for by cross-channel comparisons of envelope coherence. However, in places where the two SAM tones can interact along the basilar membrane, the resulting envelope in the anti-phasic conditions will be almost flat. The difference between coand anti-phasic modulations could then be explained by within-channel cues only. Some limited cross-channel processing had to be introduced in models in order to reproduce the low roughness of wide-band noise, but the hypothesised mechanism has not been directly tested. New stimuli were designed to address this issue. The stimuli were derived from SAM-tones by introducing a time-jitter in their envelope on a period-to-period basis. A jAM tone can be described by its average modulation frequency fm, its centre frequency fc, its amount of jitter (that controls the bandwidth), and its modulation depthm (that precisely sets the envelope rms value). All jAM tones with a same set of parameters should thus produce exactly the same roughness. It is however possible to obtain incoherent envelopes within a set of parameters by using different random samples for the jitter. Two jAM tones can then be added, without producing as important within-channel cues as anti-phasic SAM-tones.