Abstract
This paper presents a real-time implementation of the wavelet transform (WT) approach on DSP platforms for a 21-channel cochlear implant (CI) stimulation strategy. DSP platforms provide an effective cost and portable platform for cochlear implant studies. The cochlear implant converts sounds which are received from the outside, into electrical pulses that stimulate nerve endings in the cochlea, in real time. These electrical stimuli are interpreted by the brain as sounds. Different versions of our implementation are examined and compared in terms of processing time speech and accuracy. For real-time experiment, it is crucial that the speech signal be processed with minimal delay from input to output. To achieve this purpose, optimisation techniques have been taken. Thanks to these steps, we have been able to reduce the processing time to approximately 10.5 μs on a 720 MHz TMS320C6416 DSP board. This experiment provides directive guidelines for our work to treat the speech signal sample per sample at real-time operation.