The Application and Computer Simulation of Multi-Channel Cochlear Implant Based on All Phase DFT Filter

Band-pass filter bank is an essential part in cochlear implant (CI). Its characteristics are important for synthesized sound quality because the outputs of filter bank contain a lot of fine structure cues which need to be encoded and transmitted to implanted electrode and then to stimulate auditory nerves. All phase DFT (APDFT) filter is a novel and high efficient digital filter, which possesses concurrent merits, such as zero-phase (or all-phase), abrupt cut-off characteristic. This paper presents the principle and the design method of APDFT filter bank and applies this filter into CI signal processing. Under the same acoustic simulation conditions, based on the continuous interleaved sampling (CIS) strategy, we analyze and evaluate the simulation results. Comparing with classical Butterworth filter bank adopted in CI processors, some objective waveform and spectra show that the output signal from APDFT filter bank is much closer to the original sound. For the normal-hearing (NH) listener, the synthetic speech and music by the APDFT filter bank have higher quality, and perceptual tests on six NH listeners primarily verify the improvement of intelligibility in noise. Thus, the APDFT filter could potentially improve the hearing quality for CI user.