A perceptual space that can explain the robustness of bio‐acoustic communication

Abstract

The sounds that animals use to communicate, including the syllables of speech, have a very special ‘‘pulse resonance’’ form, which automatically distinguishes them from background noise. The parts of the body used to produce these sounds grow as the animal grows. Thus, there is ‘‘acoustic scale’’ variability in communication sounds, which poses a serious problem for the perception and recognition stages of communication. The success of bio‐acoustic communication suggests that the auditory system has a special preprocessor that automatically normalizes for acoustic scale as it constructs our internal ‘‘auditory image’’ of a sound. In this paper, we propose that the normalization is performed in the early stages of the auditory pathway by adding an extra, rather special, dimension to the space of auditory perception. This paper is about the mathematics of the space, which has to be ‘‘scale‐shift covariant’’ to support communication, and the discovery of a unitary operator that can construct the appropriate space. The mathematics makes it clear that there is no equivalent means of scale normalization available in the traditional time‐frequency space of the spectrogram. [Research supported by the UK MRC (G0500221, G990369).]