Thesis (M. Sc.)--University of Hong Kong, 1998. / Includes bibliographical references.

Thesis (Ph.D)--Electrical and Computer Engineering, Georgia Institute of Technology, 2009. / Committee Chair: G. Tong Zhou; Committee Co-Chair: Xiaoli Ma; Committee Member: David V. Anderson; Committee Member: James Stevenson Kenney; Committee Member: Liang Peng; Committee Member: William D. Hunt. Part of the SMARTech Electronic Thesis and Dissertation Collection.

Thesis (Ph. D.)--Drexel University, 2007. / Includes abstract and vita. Includes bibliographical references (leaf 195).

On-device audio recognition, in particular keyword spotting, will be instrumental to realizing the promise of pervasive intelligence. On-device operation demands ultra-low power and compact area. The state of the art in fully-integrated keyword spotting chips reveals that the power and area bottleneck is not the backend keyword spotting classifier, but rather the frontend audio feature extractor, motivating research into frontend audio feature extraction that is both power- and area-efficient. After, first, introducing the topic of ultra-low power audio feature extraction using time-mode analog signal processing circuits, we, second, present an analog audio feature extractor chip that achieves the lowest power/feature and area/feature, as compared, respectively, to the most area-efficient and power-efficient published analog audio feature extractor chips. Despite the chip's state-of-the-art efficiency, competitive keyword spotting accuracy is maintained when interfacing the chip with a standard, small-footprint, software backend neural network. The chip's efficiency is due to a pair of novel circuit techniques we developed. The techniques are based on time-mode analog signal processing. This is a paradigm favored by technology scaling, in which analog information is encoded in the timing of digital edges, enabling digital gates to perform analog signal processing. Third, we present a theory-based analysis of one of the two circuit techniques. Fourth, we present theory- and simulation-based progress towards what would be a novel type of analog filtering, ``Time-Mode Analog Filter." Such a filter would use only the horizontal time axis to represent and process continuous-valued information, and would be built out of nothing more than digital gates. Fifth, and finally, we present a simulation-based study that finds that in state-of-the-art analog audio feature extractor chips, the power consumption of the critical block, the analog filterbank, can be reduced by one-and-a-half orders of magnitude, while degrading downstream keyword spotting accuracy by only a couple percent, paving the way towards more rigorous system-level design of audio recognition systems.

No description available.

Requirements exist for a unique type of underwater acoustic source. The transducer is in the form of a linear array of discrete elements and is required to have a constants transmitting voltage response and carefully controlled directivity characteristics over a two octave bandwidth. A generalized model of a linear array of cylindrical piezoelectric ceramic acoustic radiators is developed and applied to the design of a prototype which operates over approximately one half of the required bandwidth. The prototype transducer was built and the measured results are compared with those predicted by the model. Recommendations are made for improving the performance of both the prototype and the array required to meet the full bandwidth specified.

This thesis is concerned with the estimation of the acoustic source signal-bearing in the presence of a reflective surface. The method to estimate the bearing of the acoustic source is the cross spectral analysis of two microphones in which the characteristics of the acoustic source signal is preserved. The echo of the original time history is removed by an advanced signal processing technique, called Cepstrum analysis. This technique is successfully applied to remove the contaminating effect of echoes in the measured time history. The work described is divided into several steps. A computer program was developed to examine the effects of reflection and the behavior of the Cepstrum under different operating conditions using simulated signals. The algorithm was then adapted to process experimental signals acquired under laboratory test conditions. An improved liftering process based on the knowledge of the echo time was developed in order to make the method relatively automated. The results of the experimental work show that the Cepstral analysis can be both effective and efficient in removing reflected signals from convolved time histories and thus successful in correcting a bearing estimate contaminated by an echo. / M.S.

This study presents a basis for the analytical and experimental procedures as well as design techniques required in achieving adaptive structures for active structural acoustic control (ASAC). Test structures studied in this work included a baffled simply supported beam and a baffled simply supported plate which were subjected to a harmonic input disturbance created physically with a shaker and modelled by a point force input. Structural acoustic control was achieved with piezoelectric actuators bonded to the surface of the test structure. The primary focus of this work was devoted to studying alternative sensing techniques in feed forward control applications. Specifically, shaped distributed structural sensors constructed from polyvinylidene fluoride (PVDF), distributed acoustic near-field sensors constructed from PVDF, and accelerometers were explored as alternatives to microphones which are typically implemented as error sensors in the cost function of the control approach. The chosen control algorithm in this study was the feed forward filtered-x version of the adaptive LMS algorithm. A much lower level of system modelling is required with this method of control in comparison to state feedback control methods. As a result, much of the structural acoustic coupling (i.e. system modelling) must be incorporated into the sensor design. / Ph. D.

Active Structural Acoustic Control (ASAC) has been previously demonstrated for systems excited by single and multiple frequency disturbances. This work is an extension of ASAC techniques to the control of sound radiation from structures excited by a disturbance with broadband frequency content. An adaptive, multi-input multi-output (MIMO), feedforward broadband acoustic control system has been developed. The control approach is the least mean squares (LMS) algorithm. The compensators are adaptive finite impulse response (FIR) digital filters. The system identification of the control loop transfer functions were implemented with infinite impulse response (IIR) digital filters. The control inputs were implemented with piezoelectric ceramic actuators (PZT). Both far-field microphones and polyvinylidene fluoride (PVDF) structural sensors designed to optimally control the efficient acoustic radiating modes were used as error sensors. The disturbance was band-limited zero mean white noise and was input with a point force shaker. In the control of harmonically excited systems, satisfactory attenuation is possible with a single-input single-output (SISO) controller. In contrast, for Systems excited with broadband disturbances, a MIMO controller is necessary for significant acoustic attenuation. Experimental results for the control of two simplySupported plates are presented. Aspects addressed include the evaluation of the microphone and PVDF error sensors, optimization of sensors and actuators, FIR compensator size, controller sample rate, and convergence time. Thus this work provides a methodology for controlling broadband acoustic radiation from a structure with regard to the practical aspects of ASAC. / Master of Science

A one dimensional normalized model for the frequency response of the acoustic power transmitted through nonuniform materials is developed. Using the ideal mixture model to relate acoustic velocity and impedance, this normalized model demonstrates that the power transmission characteristics are completely determined using only a composition profile and the parameters defining percent variation in acoustic velocity and impedance. For purposes of comparison, an analytically exact solution for exponential tapers is obtained. / Master of Science