Angle estimation and binaural processing in animal echolocation

Publisher Website
Google Scholar
Abstract
Physical structures separate the ears of many mammals, and the ears themselves have complex transfer functions that depend upon the direction of a sound source. Neither of these observations is usually incorporated into standard array analysis. The conventional array problem generally assumes that individual elements are omnidirectional, and there are no physical structures between elements to provide partial isolation. This investigation of binaural angle estimation specifically incorporated the effects of partial isolation and directivity patterns that vary with angle and frequency. Cramer-Rao bounds for the variance of maximum likelihood angle estimates were derived. These bounds, along with a generalized ambiguity function, were used to evaluate the significance of isolation between the ears, angle-dependent magnitude and phase variations in the spatial transfer functions of the ears, the use of wide-band and narrow-band signals in animal echolocation, symmetry of aural directivity patterns, the transmitted beam patterns that are employed by certain bats and the angle-dependent signal variations that have been observed in cetacean echolocation. Only 2 broadband hydrophones are sufficient for accurate, unambiguous azimuth measurement if the width of the signal autocorrelation function is small relative to the distance between hydrophones. The analysis indicates that many past measurements of transmitted beam patterns and aural transfer functions in animals are incomplete, since they do not consider direction-dependent phase variations. It may also be ncessary to reinterpret the meaning of time difference and pressure difference cues for localization. These insights can be used not only for better correlation of behavioral and physiological measurements, but also for better design of man-made systems for communication, radar, sonar and ultrasonic medical diagnosis.

This publication has 5 references indexed in Scilit: