Not available / text

Not available

The dissertation presents the design and development of a systematic signal analysis and representation framework beyond short-time Fourier power spectrum for sounds, in particular environmental sounds. This framework is consistent with the underlying assumptions of the analysis method and its elements are correlated with human perception. The sound signal has to conform to certain conditions for its power spectrum to have a physical and perceptual meaning. We contend that very few environmental sounds readily meet these criteria and argue that the quantities that are traditionally used to describe sounds need to be repurposed and, if necessary, redefined to represent sounds by non-Fourier means. We propose a perceptuo-analytic organization of sounds so that any environmental sound can be analyzed based on its signal characteristics and perception. We present environmental sound acquisition in the context of collection and annotation of a database for the footstep sounds, a common environmental sound, and show that it can be represented by these unconventional means and further analyzed to produce descriptions which are obscured with the traditional analysis. We present a novel application of extracting gait characteristics from the footstep sounds which is enabled by the proposed framework.

Active Noise Control (ANC) is an old concept which has generated increased interest over the past 10-15 years. Using the principle of destructive interference of waves, an inverse pressure wave - "anti-sound wave" is generated in order to attenuate the undesired noise. To achieve substantial cancellation of sound, performance of the cancelling sources must be accurately monitored and controlled. This has only become possible with the rapid development of digital signal processing theory and hardware. Most of the early work in the area of ANC has been done in duct silencing using single channel feed forward and feedback control arrangements. Providing that the sound wavelength is large enough (Le. frequency low enough) in comparison with the cross-sectional dimensions of the duct, spherical sound waves can be adequately approximated with plane waves. The problem is then reduced from three to two dimensions, which provides the possibilities for better understanding of the basic mechanisms of active noise cancellation and study of various adaptive control algorithms. The aim of the present work is to systematically investigate ANC methods for outdoor applications, through the development of Electronically Controlled Acoustic Shadow (ECAS) systems. In this work, the problem is fully three-dimensional. Multichannel ANC methods are proposed to be used, to reduce the noise emitted by large vibrating structures, such as power transformers, in the open air. The adopted approach is to design an active sound wall to create a controlled "anti-sound" shadow. In this way unwanted sound can be reduced in the direction of a complaint area. The potential applications for outdoor ANC systems are considerable. There is need to reduce low frequency sound, which is very hard to reduce using conventional methods - very heavy and expensive structures are required. This opens up the whole field of reducing noise from heavy rotating machinery, such as large generators/motors, factory machinery and mills (many of which have to operate 24 hours per day to remain competitive - which in turn causes noise problems). This work is divided into two main parts. First part considers computer modelling, simulations and theoretical investigation of Electronically Controlled Acoustic Shadows (ECAS) systems. It is demonstrated, that these shadows can be superior to acoustic shadows generated naturally by solid barriers. Detailed analysis predicts that deep shadows (> 1 00 dB) are po.ssible, indicating that practical shadows (>20 dB) are potentially achievable. The object of second part of the work is to investigate practical ECAS systems and establish their performance. In Chapters 2 and 3 (PART 1) the system performance at the fundamental, 100Hz frequency of transformer noise is analysed. To investigate the influence of a large number of parameters on the active wall performance, computer modelling of the primary and secondary (cancelling) sources is developed. The acoustic radiation from this primary source distribution is computed in the far field over a given control angle (both azimuthal and elevation angles). Angles between the 150 and 600 in azimuth and 150 to 300 in elevation are co~'sidered. Phase and amplitude of the secondary sources are than computed through the matrix algebra using exact solution of the least squares problem to minimise the sound at the sensor array. Using this modelling important properties of the acoustic shadows generated by active walls are established, and the basic theory to explain these shadows is formulated. No such theory existed previously. The concept of generating an acoustic shadow in the direction of the complaint area, has resulted in the acoustic properties of a 15°xI5° reference shadow being established in detail. It appears that any arbitrary shadow at this frequency can then be constructed by an addition of the~~ reference shadows, the shadow depth depending on the density of the cancellers per unit angle. Deep shadows in access of 100 dB are predicted, making practical shadows from real sources a possibility. It is now feasible to predict and optimise the future performance of proposed active wall configurations using the computer modelling and developed theory. Further. in the first part of the document (Chapter 4). acoustic interference across high frequency finite Source distributions is studied. The basic theory of non compact sources is considered and the possibility of continuous source representation with a finite number of discrete sources is discussed. The concept of non discreteness or poor discrete representation is established. Here, the .~coustic wavelength is considered small compared to the separation distance between discrete sources. The extent of the near field from these discrete source arrays is also established. where the simplified far field radiation equation breaks down. Finally, in Chapter 4. the optimisation and performance of cancelling arrays to create acoustic shadows from non compact. discrete representation of finite source distributions is investigated.

This thesis deals with some of the theoretical and practical aspects relating to the conversion gain and noise performance of mixers employing dual-gate field effect transistors (DGFET'S) . To start with, the role of mixers in the context of radio conununication receivers is highlighted and the most relevant mixer properties are explained. Solid state mixing devices and their circuits are then discussed with special emphasis on the DGFET.This includes a survey and explanation of mixing devices ,planar transmission lines and circuit components for the practical design of mixers. Chapter 3 then deals with the mixer signal analysis as well as accuracy considerations. A more detailed treatment of the DGFET in terms of structure and dc model is given in the subsequent chapter. Problems associated with the choice of an FET model are referred to as well as the use of MATLAB for computationa: purposes. This is followed in the next two chapters with the development and analysis of the large signal equivalent circuit of tr.{· DGFET,and a treatment of noise and its measurement as associated with mixers. The design, practical implementation and measurement of the properties of DGFET mixers is covered in chapter 7. This is followed in Chapter 8 by an overall discussion of results, possible future work and conclusions. A new FET model is proposed that enables the dc characteristics to be simulated more closely than in previous models, particularly at low drain voltages. Furthermore, the representation of the noise by a frequency independent drain current generator and an input noise conductance enabled a single set of measurements to simulate the noise behaviour of the device as an amplifier or a mixer. Practical investigations using an NEe device type NE 41137 gave a maximum stable conversion gain in the frequency range O.SGHz to 3.0GHz of 4dB with a minimum noise figure 8.SdB.

published_or_final_version / abstract / toc / Orthopaedics and Traumatology / Master / Master of Philosophy

published_or_final_version / Computer Science / Doctoral / Doctor of Philosophy

No description available.

No description available.

This work is devoted to the speech signal clustering analysis problem. The main methods of cluster analysis were reviewed in this work and clusterization algorithm based on vector quantization was created. The speaker identification experiments were performed in which dependence of identification accuracy and computational complexity from number of clusters was investigated.