Semester
Summer
Date of Graduation
2009
Document Type
Thesis
Degree Type
MS
College
Statler College of Engineering and Mineral Resources
Department
Lane Department of Computer Science and Electrical Engineering
Committee Chair
David W. Graham.
Abstract
The objective of this work is to develop analog integrated circuits to serve as low-power auditory front-ends in signal processing systems. An analog front-end can be used for feature-extraction to reduce the requirements of the digital back-end, or to detect and call attention to compelling characteristics of a signal while the back-end is in sleep mode. Such a front-end should be advantageous for speech recognition, noise suppression, auditory scene analysis, hearing prostheses, biological modeling, or hardware-based event detection.;This work presents a spectral decomposition system, which consists of a bandpass filter bank with sub-band magnitude detection. The bandpass filter is low-power and each channel can be individually programmed for different quality factors and passband gains. The novel magnitude detector has a 68 decibel dynamic range, excellent tracking capability, and consumes less than a microwatt of power. The system, which was fabricated in a 0.18 micron process, consists of a 16-channel filter bank and a variety of sub-band computational elements.
Recommended Citation
Rumberg, Brandon David, "Speech Processing Front-end in Low-power Hardware" (2009). Graduate Theses, Dissertations, and Problem Reports. 2830.
https://researchrepository.wvu.edu/etd/2830