There I mainly worked on designing a MP3 player based on a LuxSonor IC product LS800 (a multimedia processor with embedded real-time operating system); I also wrote drivers for audio interface and remote control. In the fall of 2000, I began my Ph.D. study in Dr. Kwabena Boahen's lab at University of Pennsylvania. Since then I have been working on morphing the cochlea, the front-end of audition.
The cochlea turns the sound pressure into mechanical vibrations, generating neuron spikes transmitted all the way up to the auditory cortex for further processing. The cochlea does both spectrum and spatial analysis, responding to high and low frequencies at basal and apical locations along its spiral structure. Complex and delicate, the cochlea exhibits remarkable sensitivity, selectivity, and amplification when processing frequencies and intensities. Enormous physiological measurements of the cochlear responses to sound stimuli have provided quite much evidence for revealing part of the cochlear mechanics. However, due to the vulnerability of the cochlea, physiological data themselves cannot tell a complete story in detail about how the cochlea works. Thus, numerous mathematical cochlear models have been developed to match the continuously improving biological data, and to help uncover the mystery. Recently silicon cochlea models have also been built to emulate the cochlear behavior. These analog Very-Large-Scale-Integration (VLSI) circuits offer the advantage of real-time computation (over software simulations), accessibility for measurements (compared to their biological counterparts), and comparable computation efficiency, occupying space, and power consumption to the biology. My research goal is to verify the hypothesized cochlear mechanics through building both a mathematical model and an analog VLSI circuit model.
I have built a mathematical model that simulates the nonlinear active cochlear responses. Extending Geisler and Sang's model , the present model includes the tilt of both the outer hair cells (OHCs) and the phalangeal processes (PhPs), which generate active bi-directional coupling OHC forces. Then following Watts's approach , I have explored a semi-analytical solution to a linear version of the model in order to obtain more insights into the active amplification. And now I am testing the cochlear chip I designed early this year.
Simulation results in the frequency domain in MathematicaTM of our 2D linear active model at different tone input . For each case, top figure shows the pressure difference of the cochlear fluid across the BM while the bottom one shows the BM displacement with time.
2k Hz pure tone
0.5k Hz pure tone