Investigation and Evaluation of Low-Cost, Low-Power Mixed Signal Auditory Processing Systems

本研究的目的是探索基于vlsi的模拟前端与基于dsp的后端混合处理听觉信号的潜力。基于这项研究的结论和进展，预计我们将能够确定未来的研究路径，特别是针对将基于模拟vlsi的预处理结合到现有听觉处理系统的产品的商业化。当然，这项研究的结论可能表明，将模拟VLSI与DSP结合用于这些应用在一定程度上是徒劳的。然而，即使在这种最坏的情况下，我们也希望从这项研究中获得对模拟vlsi电路在听觉和其他非视觉处理应用中的潜力的宝贵见解。我们提出的探索性研究分为：理解和应用特定的DSP架构（Analog Devices 2100系列）来解决简单的听觉处理问题；改进模拟VLSI中紧凑、高度可调的带通滤波器的设计和实现，以适应DSP的优势。结合DSP和改进的模拟VLSI前端处理简单的听觉信号，比较模拟VLSI、DSP和混合信号系统在处理这些相同的听觉信号时，在功耗、空间、预期成本和识别精度方面的性能。＊＊＊

9907464WilsonThe purpose of the proposed research is to explore the potential for mixing analog VLSI-based front-ends with DSP-based back-ends for processing auditory signals. Based on the conclusions and progress of this research, it is expected that we will be able to determine future paths of research, especially directed toward commercialization of products that incorporate analog VLSI-based pre-processing onto existing auditory processing systems. It is certainly possible that conclusions of this research might suggest a degree of futility in combining analog VLSI with DSP for these applications. Even in this worst case sce-nario, however, we expect that valuable insight into the potential of analog-VLSI circuits for auditory and other non-visual processing applications will be gained from the research.The exploratory research we propose is divided into:Understanding and applying a particular DSP Architecture (Analog Devices 2100 series) to simple auditory processing problemsImproving the design and implementation of compact, highly tunable bandpass filters in analog VLSI to suit the strengths of the DSP.Combining the DSP and improved analog VLSI front-end to process simple auditory signalsComparing the performance of analog VLSI, DSP, and mixed signal systems in processing these same auditory signals in terms of power, space, anticipated cost, and recognition accuracy. ***