Two speech processing schemes for the University of Melbourne multi-channel cochlear implant prosthesis

A series of speech perceptual studies was conducted on normally hearing subjects using synthesized acoustic signals based on two speech processing schemes: zero-crossing and filter bank. The filter bank scheme was shown to provide more speech information than the zero-crossing scheme. These schemes were implanted in a laboratory-based speech processor, which is described. A low-power switched capacitor speech spectrum analyzer embodying several novel design methodologies is also described. This spectrum analyzer can be used to implement the filter bank scheme in a wearable speech processor.<
>