Date of Graduation
Statler College of Engineering and Mineral Resources
Lane Department of Computer Science and Electrical Engineering
David W Graham
The need for efficient electronics has been increasing by the day, as have the constraints on power and size of the devices. Also the increase in use of mobile and wearable electronics has been leading to innovative methods to conserve power and increase functionality. The traditional approach of signal processing heavily relies on the Digital Signal Processing (DSP) hardware to perform most of the tasks, which has lead to power-hungry circuits. Use of analog front-end devices could prove to be efficient, since most of the real-world data is analog and since the DSP could be spared for more application-specific tasks within the system, thereby resulting in more efficient mixed-signal systems.;The focus in this work is to develop an analog front-end for speech-processing applications with inspiration from biology, and trying to mimic human auditory perception techniques. The circuits are designed in 600nm, 350nm and 180nm CMOS processes and are biased in the sub-threshold region to consume low-power. Also, various modules of the system are connected using multiplexing circuits to allow post-fabrication reconfigurability to suit various applications. These circuits are biased using a network of floating-gate transistors which allow reconfigurability and increased bias accuracy. This thesis mainly describes two modules of the analog front-end used for speech processing: derivative circuit and voltage-mode subtractor circuit, which are used for processing spectrally decomposed signals. These circuits could be used for applications like audio analysis or event detection.
Singireddy, Anvesh Kumar, "Low-Power Analog Circuits for Sub-Band Speech Processing" (2010). Graduate Theses, Dissertations, and Problem Reports. 4657.