Coupling-induced tunability of characteristic frequency, bandwidth and gain of artificial hair cells

Abstract

Drawing inspiration from nature, we develop bio-inspired acoustic sensors with integrated signal processing capabilities to (i) close the performance gap between the human hearing and machine hearing and (ii) test models on biological hearing. Particularly important is thereby the combination of frequency decomposition with nonlinear (compressive) amplification of the sound signals. Here, the question arises, how the frequency resolution of 0.1–0.4%, the large gain and the coverage of the large frequency range of 20 Hz to 20 kHz can be obtained with a modest number of 3000 inner hair cells as transducers without missing tones. To solve this issue, it was hypothesized that the cochlea can be modeled as coupled critical oscillators. We study experimentally and theoretically the effects of coupling critical oscillators using bio-inspired acoustic sensors, which are based-on microelectromechanical system (MEMS) resonators with a high-quality factor and a resonance frequency set by the geometry. Using electronic feedback, these resonators act like critical oscillators tuned near Andronov–Hopf bifurcation point. If output-signal coupling is added, three different bifurcation points are generated. Tuning the system close to one of these bifurcation points leads to a highly tunable behavior and sound pressure dependent sensitivity that is compressive in nature. In this case, the response frequency of the sensor system can be shifted by tuning the control parameter for bifurcation, allowing to cover larger bandwidths with one sensor pair while retaining high quality factors. Furthermore, tuning coupling and feedback strength, bandwidth and gain of each sensor can be adapted as needed. Using these effects, an adaptive filter bank to model the cochlear functionality and adaptation can be build. Since efferent feedback can tune the response of outer hair cells and thus inner hair cells and basilar membrane as well, the question arises if such tuning mechanisms can be observed in the mammalian cochlea as well.