Silicon models of the auditory pathway

![CDATA[This chapter introduced neuromorphic engineering, and a number of different silicon models that represent various functions in the auditory pathway have been presented. As understanding of the auditory pathway has improved, the VLSI designs have utilized models that have gradually included more and more of the impressive signal processing characteristics that are found in biology. These silicon models are used in a variety of applications, from a tool for researchers to validate mathematical models in real time to audio front-ends in speech recognition systems. The complexity of a model used will depend on its end application, however, the spirit of neuromorphic engineering remains the same no matter what the application: to leverage biological functionality in modern technology. The quest to build better and more realistic silicon auditory pathway models is being confronted with many of the challenges that biology has had to overcome. While there has not been a silicon model built to date that rivals biology, the situation is improving, with better models and better technology. In just over 20 years, implementations have improved from simple 1-D filter cascades in silicon cochlea modeling to more biologically plausible 2-D structures that include outer hair cell functionality. There have also been improvements in circuit design and as miniaturization of CMOS fabrication technology continues, we can include more circuits (filters, cell structures, etc.) on a single integrated circuit. Even after two decades since Lyon and Mead’s (1988) first silicon cochlea, designers are still faced with many of the same design challenges: noise, dynamic range and mismatch to name but a few. This chapter has presented some examples, however, to show that silicon designs have improved significantly in 20 years and by overcoming some of the design challenges that still face the neuromorphic engineer.]]