Several audiological tests require knowledge of the sound-pressure spectrum at the eardrum. However, microphone readings are typically made at another, more-accessible position in the auditory canal. Recordings are then "adjusted" to the plane of the eardrum via mathematical models of the ear canal and eardrum. As bandwidths of audiological instruments have increased, ear-canal models have, by necessity, become more precise geometrically. Reported herein is a noninvasive procedure for acquiring geometry of the ear canal in fine detail. The method employs a computer-assisted tomographic (CAT) scanner in two steps to make radiographic images of parasagittal cross sections at uniform intervals along the lateral length of the canal. Accuracy was evaluated by comparing areas of cross sections appearing in radiographic images of a cadaver ear canal to cross sectional areas of corresponding michrotome slices of an injection mold of the same canal. Percent differences between these two areas had a mean value of 9.65% for 26 different cross sections of the one ear canal studied. Ear canal volume estimated from the CAT images was 6.12% different from the estimated volume of the injection mold: an improvement over the reported 39% maximum error of conventional acoustic volume measurements.

Mathematical modeling of acoustical systems is a popular design tool. To verify math models for acoustical systems of small physical size, probe-tube microphones are often used. Unfortunately, the presence of the probe tube, itself, sometimes affects the sound field being measured. One way to solve the problem is to account for this effect by incorporating a model of the probe tube into an overall math model of the entire system. Equations for analyzing sound transmission through probe tubes are not available in most contemporary acoustics textbooks. This paper describes a method for mathematically modeling the acoustical characteristics of probe-tube microphones using equations developed in 1950 for application to transient fluid flow in pipes. The accuracy of the method is demonstrated by mathematically simulating a ' probe-tube calibration experiment in the laboratory. The experimental data and data derived from the math model agree quite well over the frequency range from 10 to 10000 Hz. This shows that the math model of the probe-tube microphone used in this experiment was correct. The result also suggests that this model may be included in an overall system model, given that several restrictions listed in this paper are closely followed.

Ear simulators were designed to replicate acoustical characteristics of the average adult ear. Due to variability of ear-canal geometry and eardrum impedance among individuals, the possibility of any one person exhibiting such "average" characteristics--especially if that person is a child and/or has a conductive pathology--is remote. Thus, ear simulators have been of only peripheral value when prescribing a hearing aid (a high output impedance device) to fit the acoustical requirements of a particular patient. Reported herein is development of a programmable artificial ear (PAE) that can account for individual differences in ear-canal geometry and eardrum impedance. It consists of a 2.0-cc coupler, microphone, amplifier, computer, PAE code, and a computer card and/or software for digitization and Fourier transformation. Required input data includes ear-canal dimensions, eardrum impedance, and output impedance of the hearing aid being tested. Sound-pressure recordings produced in the 2.0-cc coupler by the hearing aid are adjusted by the computer to what they would have been had the recordings been made at the eardrum of a particular patient wearing the same hearing aid. Good agreement was observed between experiment and theory for one test case involving a totally occluding miniature earphone.

Suppressing unstable acoustic feedback in hearing aids will first require knowledge of the open-loop transfer functions of such systems. Reported herein is a mathematical technique for simulating the open-loop transfer function of an in situ eyeglass-type hearing aid. In particular, a computer program was developed that characterized the hearing aid as a serial connection of two-port blocks, each representing one individual component of a hearing aid. Included, for example, were two-port blocks representing the microphone, amplifier, receiver, sound tubes leading to the eardrum (including the ear canal itself), earmold vent, and external pathway from the vent outlet back to the microphone. The computer program was validated by replicating laboratory data derived from an experiment involving a nonstandard manikin fitted with a nonstandard artificial ear. Next, the open-loop transfer function of an eyeglass-type hearing aid in situ on the manikin was simulated via the computer program. Unfortunately, those computer-generated data were not replicated in the laboratory due to the difficulty encountered in actually measuring the open-loop transfer function. Nevertheless, investigators were able to utilize those data to predict, within +/- 25 Hz, the "squeal" frequency of unstable acoustic feedback.

Acoustic feedback in hearing aids has received little attention in the literature. Feedback occurs when stability conditions of the open-loop transfer function of an in situ hearing aid are violated. Solving the feedback problem will first require knowledge of the open-loop transfer function. Included in the open-loop transfer function is the acoustical path by which sound emanating from the earmold vent returns to the microphone (i.e., the feedback path). Reported herein are two different mathematical procedures for simulating transfer functions of the feedback path of an eyeglass-type hearing aid. In one procedure the vent exit was modeled as a point source of sound located on a flat plane, while it was treated as a point source on a sphere in the other. Results of laboratory experiments indicate that the mathematical models accurately predict those acoustic phenomena for which they were intended: point sources on plane and spherical baffles. Results of manikin experiments showed both models to be less accurate for simulating the feedback path around the human head. The maximum difference between experiment and theory was 6 dB at one frequency. Surprisingly, the flat-baffle model produced better agreement with experimental results than did the sphere model.