ABSTRACT
In the auditory image model, the tone chroma of the sound is determined with the aid of a “spiral processor” (Patterson, 1986; Patterson & Holdsworth, 1990). In essence, the processor looks at the activity on sets of vertical slices of the auditory image that are separated by doublings in time to see if one such set has more activity than the others (Patterson & Nimmo-Smith, 1986). For the sounds in the current experiment, the sequence of vertical slices at 1, 2, 4, 8, 16, 32 and 64 ms is correctly identified as the tone chroma in every case. The tone height of the notes is not specified by the spiral processor in its original form. But the fact that the octave-response data are closely related to the activity on the 4-and 8-ms slices of the auditory images suggests that the spiral algorithm might be extended to include a measure of tone height, and so become a complete model of musical pitch. Accordingly, the model was used to extract slices through the auditory images in Figure 3 at 1 ms and its successive doublings. The slices at the 8-ms point and the 4-ms point are presented in the left and right columns of Figure 4, respectively. The abscissa is channel number; the spectal resolution was increased to 99 channels for these plots. For convenience, all of these activity patterns will be referred to as ‘spiral excitation patterns’, or simply ‘excitation patterns’ when there is no ambiguity.
