Spectral coding methods for speech compression and speaker identification

This thesis investigates aspects of encoding the speech spectrum at low bit rates, with extensions to the effect of such coding on automatic speaker identification. Vector quantization (VQ) is a technique for jointly quantizing a block of samples at once, in order to reduce the bit rate of a coding system. The major drawback in using VQ is the complexity of the encoder. Recent research has indicated the potential applicability of the VQ method to speech when product code vector quantization (PCVQ) techniques are utilized. The focus of this research is the efficient representation, calculation and utilization of the speech model as stored in the PCVQ codebook. In this thesis, several VQ approaches are evaluated, and the efficacy of two training algorithms is compared experimentally. It is then shown that these productcode vector quantization algorithms may be augmented with lossless compression algorithms, thus yielding an improved overall compression rate. An approach using a statistical model for the vector codebook indices for subsequent lossless compression is introduced. This coupling of lossy compression and lossless compression enables further compression gain. It is demonstrated that this approach is able to reduce the bit rate requirement from the current 24 bits per 20 millisecond frame to below 20, using a standard spectral distortion metric for comparison. Several fast-search VQ methods for use in speech spectrum coding have been evaluated. The usefulness of fast-search algorithms is highly dependent upon the source characteristics and, although previous research has been undertaken for coding of images using VQ codebooks trained with the source samples directly, the product-code structured codebooks for speech spectrum quantization place new constraints on the search methodology. The second major focus of the research is an investigation of the effect of lowrate spectral compression methods on the task of automatic speaker identification. The motivation for this aspect of the research arose from a need to simultaneously preserve the speech quality and intelligibility and to provide for machine-based automatic speaker recognition using the compressed speech. This is important because there are several emerging applications of speaker identification where compressed speech is involved. Examples include mobile communications where the speech has been highly compressed, or where a database of speech material has been assembled and stored in compressed form. Although these two application areas have the same objective - that of maximizing the identification rate - the starting points are quite different. On the one hand, the speech material used for training the identification algorithm may or may not be available in compressed form. On the other hand, the new test material on which identification is to be based may only be available in compressed form. Using the spectral parameters which have been stored in compressed form, two main classes of speaker identification algorithm are examined. Some studies have been conducted in the past on bandwidth-limited speaker identification, but the use of short-term spectral compression deserves separate investigation. Combining the major aspects of the research, some important design guidelines for the construction of an identification model when based on the use of compressed speech are put forward.

Coding speech signals at very low rates (below 1 kb/s) with high intelligibility

This thesis presents an original approach to parametric speech coding at rates below 1 kbitsjsec, primarily for speech storage applications. Essential processes considered in this research encompass efficient characterization of evolutionary configuration of vocal tract to follow phonemic features with high fidelity, representation of speech excitation using minimal parameters with minor degradation in naturalness of synthesized speech, and finally, quantization of resulting parameters at the nominated rates. For encoding speech spectral features, a new method relying on Temporal Decomposition (TD) is developed which efficiently compresses spectral information through interpolation between most steady points over time trajectories of spectral parameters using a new basis function. The compression ratio provided by the method is independent of the updating rate of the feature vectors, hence allows high resolution in tracking significant temporal variations of speech formants with no effect on the spectral data rate. Accordingly, regardless of the quantization technique employed, the method yields a high compression ratio without sacrificing speech intelligibility. Several new techniques for improving performance of the interpolation of spectral parameters through phonetically-based analysis are proposed and implemented in this research, comprising event approximated TD, near-optimal shaping event approximating functions, efficient speech parametrization for TD on the basis of an extensive investigation originally reported in this thesis, and a hierarchical error minimization algorithm for decomposition of feature parameters which significantly reduces the complexity of the interpolation process. Speech excitation in this work is characterized based on a novel Multi-Band Excitation paradigm which accurately determines the harmonic structure in the LPC (linear predictive coding) residual spectra, within individual bands, using the concept 11 of Instantaneous Frequency (IF) estimation in frequency domain. The model yields aneffective two-band approximation to excitation and computes pitch and voicing with high accuracy as well. New methods for interpolative coding of pitch and gain contours are also developed in this thesis. For pitch, relying on the correlation between phonetic evolution and pitch variations during voiced speech segments, TD is employed to interpolate the pitch contour between critical points introduced by event centroids. This compresses pitch contour in the ratio of about 1/10 with negligible error. To approximate gain contour, a set of uniformly-distributed Gaussian event-like functions is used which reduces the amount of gain information to about 1/6 with acceptable accuracy. The thesis also addresses a new quantization method applied to spectral features on the basis of statistical properties and spectral sensitivity of spectral parameters extracted from TD-based analysis. The experimental results show that good quality speech, comparable to that of conventional coders at rates over 2 kbits/sec, can be achieved at rates 650-990 bits/sec.

A process for the management of physical infrastructure

Physical infrastructure assets are important components of our society and our economy. They are usually designed to last for many years, are expected to be heavily used during their lifetime, carry considerable load, and are exposed to the natural environment. They are also normally major structures, and therefore present a heavy investment, requiring constant management over their life cycle to ensure that they perform as required by their owners and users. Given a complex and varied infrastructure life cycle, constraints on available resources, and continuing requirements for effectiveness and efficiency, good management of infrastructure is important. While there is often no one best management approach, the choice of options is improved by better identification and analysis of the issues, by the ability to prioritise objectives, and by a scientific approach to the analysis process. The abilities to better understand the effect of inputs in the infrastructure life cycle on results, to minimise uncertainty, and to better evaluate the effect of decisions in a complex environment, are important in allocating scarce resources and making sound decisions. Through the development of an infrastructure management modelling and analysis methodology, this thesis provides a process that assists the infrastructure manager in the analysis, prioritisation and decision making process. This is achieved through the use of practical, relatively simple tools, integrated in a modular flexible framework that aims to provide an understanding of the interactions and issues in the infrastructure management process. The methodology uses a combination of flowcharting and analysis techniques. It first charts the infrastructure management process and its underlying infrastructure life cycle through the time interaction diagram, a graphical flowcharting methodology that is an extension of methodologies for modelling data flows in information systems. This process divides the infrastructure management process over time into self contained modules that are based on a particular set of activities, the information flows between which are defined by the interfaces and relationships between them. The modular approach also permits more detailed analysis, or aggregation, as the case may be. It also forms the basis of ext~nding the infrastructure modelling and analysis process to infrastructure networks, through using individual infrastructure assets and their related projects as the basis of the network analysis process. It is recognised that the infrastructure manager is required to meet, and balance, a number of different objectives, and therefore a number of high level outcome goals for the infrastructure management process have been developed, based on common purpose or measurement scales. These goals form the basis of classifYing the larger set of multiple objectives for analysis purposes. A two stage approach that rationalises then weights objectives, using a paired comparison process, ensures that the objectives required to be met are both kept to the minimum number required and are fairly weighted. Qualitative variables are incorporated into the weighting and scoring process, utility functions being proposed where there is risk, or a trade-off situation applies. Variability is considered important in the infrastructure life cycle, the approach used being based on analytical principles but incorporating randomness in variables where required. The modular design of the process permits alternative processes to be used within particular modules, if this is considered a more appropriate way of analysis, provided boundary conditions and requirements for linkages to other modules, are met. Development and use of the methodology has highlighted a number of infrastructure life cycle issues, including data and information aspects, and consequences of change over the life cycle, as well as variability and the other matters discussed above. It has also highlighted the requirement to use judgment where required, and for organisations that own and manage infrastructure to retain intellectual knowledge regarding that infrastructure. It is considered that the methodology discussed in this thesis, which to the author's knowledge has not been developed elsewhere, may be used for the analysis of alternatives, planning, prioritisation of a number of projects, and identification of the principal issues in the infrastructure life cycle.