The aim of source coding is to represent information as accurately as possible using as few bits as possible and in order to do so redundancy from the source needs to be removed. The aim of channel coding is in some sense the contrary, namely to introduce redundancy that can be exploited to protect the information when being transmitted over a nonideal channel. Combining these two techniques leads to the area of joint source-channel coding which in general makes it possible to achieve a better performance when designing a communication system than in the case when source and channel codes are designed separately. In this thesis four particular areas in joint source-channel coding are studied: analog (i.e. continuous) bandwidth expansion, distributed source coding over noisy channels, multiple description coding (MDC) and soft decoding. A general analog bandwidth expansion code based on orthogonal polynomials is proposed and analyzed. The code has a performance comparable with other existing schemes. However, the code is more general in the sense that it is implementable for a larger number of source distributions. The problem of distributed source coding over noisy channels is studied. Two schemes are proposed and analyzed for this problem which both work on a sample by sample basis. The first code is based on scalar quantization optimized for a certain channel characteristics. The second code is nonlinear and analog. Two new MDC schemes are proposed and investigated. The first is based on sorting a frame of samples and transmitting, as side-information/redundancy, an index that describes the resulting permutation. In case that some of the transmitted descriptors are lost during transmission this side information (if received) can be used to estimate the lost descriptors based on the received ones. The second scheme uses permutation codes to produce different descriptions of a block of source data. These descriptions can be used jointly to estimate the original source data. Finally, also the MDC method multiple description coding using pairwise correlating transforms as introduced by Wang et al. is studied. A modi fication of the quantization in this method is proposed which yields a performance gain. A well known result in joint source-channel coding is that the performance of a communication system can be improved by using soft decoding of the channel output at the cost of a higher decoding complexity. An alternative to this is to quantize the soft information and store the pre-calculated soft decision values in a lookup table. In this thesis we propose new methods for quantizing soft channel information, to be used in conjunction with soft-decision source decoding. The issue on how to best construct finite-bandwidth representations of soft information is also studied. / QC 20100920

The objective of the proposed research is to develop a probabilistic model of speech production that exploits the multiplicity of mapping between the vocal tract area functions (VTAF) and speech spectra. Two thrusts are developed. In the first, a latent variable model that captures uncertainty in estimating the VTAF from speech data is investigated. The latent variable model uses this uncertainty to generate many-to-one mapping between observations of the VTAF and speech spectra. The second uses the probabilistic model of speech production to improve the performance of traditional speech algorithms, such as enhancement, acoustic model adaptation, etc. In this thesis, we propose to model the process of speech production with a probability map. This proposed model treats speech production as a probabilistic process with many-to-one mapping between VTAF and speech spectra. The thesis not only outlines a statistical framework to generate and train these probabilistic models from speech, but also demonstrates its power and flexibility with such applications as enhancing speech from both perceptual and recognition perspectives.

La récupération d'énergie ambiante permet d’alimenter de manière autonome des systèmes de petite taille tels que des neouds de capteurs ou des objets connectés à internet (IoT) en remplacement des batteries. Les sources d’énergie ambiante sont par exemple, l’énergie solaire, le gradient thermique, les forces mécaniques, le rayonnement électromagnétique et la pile microbienne. Les matériaux piézoélectriques permettent de valoriser électriquement l’énergie mécanique de vibration en la convertissant directement en énergie électrique. Les niveaux de puissance assez faible (de quelques μW au mW) ont amené à développer des interfaces électriques de récupération afin d’extraire le maximum d'énergie en améliorant le couplage électromécanique. Dans ce travail, nous nous intéressons à l’amélioration de dispositif de récupération d’énergie. Deux aspects sont abordés : dans un premier temps l’étude d’un commutateur hybride synchrone électrique-mécanique est faite pour remplacer le transistor MOSFET couramment utilisé, afin de réduire sa consommation d’énergie ; dans un deuxième temps, un travail est mené sur une nouvelle structure mécanique à base de poutres reliées entre elle par des forces de répulsion magnétique. La structure obtenue par cet ensemble de poutres et de type non-linéaire à plusieurs degrés de liberté (MDOF) ce qui permet augmenter la bande passante. / The future trend of Internet of Things (IoT) is bringing energy harvesting in to the core technique due to its requirement of self-power supplying. For best customer interface and eco-friendly issues, additional sensing systems are to be designed small, wireless and self-powering. Energy harvesting provides a way to realize the wireless self-powered system, it enables the device itself to obtain its own energy from their environment. Solar energy, thermal gradient, mechanical forces, are some commonly seen methods to obtain energy from the environment. The piezoelectric energy harvester is chosen to harvest vibrational energy in this study. In this work, a simple model of the original electrical smart switch driven under ultra-low power is proposed. By using the miniature device to drive the smart switch, the efficiency when low power is provided was examined. To construct an energy harvesting system in a more complete aspect, two newly proposed methods are as below: First, the hybrid-electrical-mechanical switches were utilized to replace the commonly seen electrical smart switches, to reduce its energy consumption such as threshold loss. Secondly, we designed a new mechanical structure for the cantilever array by connecting the beams using magnetic repelling force. In this way, the beams within the array were connected physically, forming a nonlinear multi-degree of freedom (MDOF) -like result.