イルワンシャー イルワンシャー

This study aimed to address the issue of "crosstalk" in bone conduction hearing aids, where sound meant for one ear is mistakenly perceived by the other ear via bone conduction. We explored a potential solution by canceling the crosstalk sound at the cochlea. To achieve this, an accelerometer was placed on the mastoid to monitor the crosstalk sound produced by an audiometric bone transducer on the opposite side of the head, while a second transducer on the same side as the accelerometer was used to cancel it out. The filtered-x least mean square (FxLMS) algorithm was used to optimize the crosstalk compensation (CTC) filter for cancellation at the mastoid. Then, the subjects manually adjusted the filter coefficients through a lateralization task to achieve crosstalk cancellation at the cochlea. This task involved modifying phase and level differences between pure-tone sounds from the two transducers, making the sound seem to originate from the leftmost or rightmost side of the head. Our results indicated successful cancellation of crosstalk sound at the cochlea, as subjects' hearing thresholds under noise masking were lower with crosstalk cancellation.

In bone-conducted sound reproduction, crosstalk is a phenomenon in which sound presented on either side of the head reaches the cochlea in the two ears. However, since the brain compares signals coming from each cochlea, crosstalk at high levels may limit the ability to hear with two ears, especially in patients with two bone-conduction hearing aids (BCHAs). In this study, we intended to suppress crosstalk sounds at the cochlea by canceling them at a sensor in the ear canal. Two sensors were evaluated here: (i) a probe microphone that can record the ear-canal sound pressure and (ii) an accelerometer that can capture the vibration of the inner wall of the ear canal. Since crosstalk cancellation was designed to occur at the sensor location, we used the tone reception threshold (TRT) as a subjective indicator to investigate which sensors contributed the most to the cancellation at the cochlea.

Having a bone conduction (BC) transducer anywhere on the head will result in a BC sound reaching the cochlea in both ears. This “cross-talk” phenomenon may limit a listener's ability with a pair of BC transducer to sense sound direction. In this paper, we discuss a way to minimize cross-talks in BC sound reproduction using a method called “cross-talk cancellation.” Ideally, the method requires transfer functions (TFs) from each of the BC transducers to each of the listener's cochleae to accurately synthesize cross-talk compensation (CTC) filters; however, a direct measurement of the TFs at the cochlea is not possible. Since the ear canal is the closest to the cochlea, we thus hypothesize that the TFs measured at the ear canal might be used to achieve cross-talk cancellation at the cochlea in the inner ear. Therefore, we utilized a BC ear microphone to capture the vibration of the bony ear-canal caused by vibrating transducers on the mastoid. The filtered-x least mean square (FxLMS) algorithm was then used to estimate the CTC filter. Experiments with and without cross-talk cancellation were done to determine the effective frequency range that could achieve cancellation in the ear canal of three normal-hearing participants.

A back-to-back speaker system can be used to achieve a cardioid-like radiation pattern in a low-frequency range. The pattern is obtained by placing two closely spaced speakers mounted back to back with a filter. The arrangement results in two sound waves that are in opposite phases, the sound wave from the back speaker cancels the back side sound wave from the front speaker. The arrangement achieves sound cancellation only in a narrow coverage angle. On the other hand, adding two more speakers may widen the coverage angle of the cancellation. This paper reports on the result of our investigation into a speaker system that consists of a primary speaker and three secondary speakers. The primary speaker is set to point in the desired direction, and three secondary speakers with semi-circular arrangement are used to reduce the back side sound wave at low frequencies where corresponding filters of each secondary speaker create the cancellation signals as a result of the filtered-x least mean square (FXLMS) Algorithm. The system has been tested and examined with measurements in an anechoic chamber. The results obtained were also compared to the results of the back-to-back speaker system.