Theses and Dissertations

Permanent URI for this collectionhttp://ir.daiict.ac.in/handle/123456789/1

Browse

Has files: YesSubject: search.filters.subject.Statistics

Search Results

Now showing 1 - 4 of 4
  • Thumbnail Image
    ItemOpen Access
    Analysis of nonlinearity in speech production mechanism for speaker verification: phase-based approach
    (Dhirubhai Ambani Institute of Information and Communication Technology, 2015) Agrawal, Purvi; Patil, Hemant A.
    Many of the real-world signal processing problems can be described using linear models, and can be realized as analog or digital filter, time-invariant filters; finite or infinite impulse response (IIR or FIR) filters. In the recent past, a nonlinear operator called Teager Energy Operator (TEO) has been introduced and investigated as it has a small window in temporal-domain, making it ideal for local time analysis of signals. This thesis aims to explore the nonlinear nature of the speech production mechanism of a speaker. There has been significant advancement in exploring the source and system-based features for speaker recognition attributed to the characteristics of the excitation source and size and shape of the vocal tract. In this work, TEO phase features are derived from fullband speech signal and then on subband speech signal (due to the property of the TEO being a monocomponent operator). In addition, a feature set is derived from residual phase extracted from nonlinear filter designed using Volterra-Weiner (VW) series exploiting higher-order linear as well as nonlinear relationships hidden in the sequence of samples of speech signal. Experiments have been performed on the score-level fusion of the proposed feature sets with state-of-the-art MFCC features for text-independent Speaker Verification (SV) task, based on Gaussian Mixture Model-Universal Background Model (GMM-UBM) system, respectively. The performance of each feature set is evaluated and a comparative study of each of the features is presented. The results obtained provide an evaluation of the nature of the speech production mechanism and provides features to improve performance of SV system.
  • Thumbnail Image
    ItemOpen Access
    Wideband active mixer with high gain and high linearity
    (Dhirubhai Ambani Institute of Information and Communication Technology, 2015) Pandey, Vijay Raghav; Gupta, Sanjeev
    An active mixer is presented with improved conversion gain and linearity over a wide range of frequency. The mixer is combined with a low noise amplifier which not only provides braodband input matching but also cancels out noise at the output. The LNA has two stages, input matching stage and noise cancellation stage. The former provides matching for a wide range of frequency and the latter cancels out the noise of the former stage at the output. A PMOS is used to cancel out the non-linear effects of the noise cancellation stage of the LNA thereby improving the linearity of the system. The non-linear effects of input matching stage is canceled by the noise cancellation stage itself. A current bleeding circuit is used to fulfill the large current requirement of the noise cancellation stage and helps in further improvement of the gain. Gilbert cell topology is used which has differential output thereby providing better immunity from noise and fluctuations. The circuit provides a conversion gain of 20 to 24 dB, noise figure of 8 to 13 dB and a linearity of 17.5 dBm for a frequency range of 1 to 6.5 GHz.
  • Thumbnail Image
    ItemOpen Access
    Learning to rank: using Bayesian networks
    (Dhirubhai Ambani Institute of Information and Communication Technology, 2011) Gupta, Parth; Mjumder, Prasenjit; Mitra, Suman K.
    Ranking is one of the key components of an Information Retrieval system. Recently supervised learning is involved for learning the ranking function and is called 'Learning to Rank' collectively. In this study we present one approach to solve this problem. We intend to test this problem in di erent stochastic environment and hence we choose to use Bayesian Networks for machine learning. This work also involves experimentation results on standard learning to rank dataset `Letor4.0'[6]. We call our approach as BayesNetRank. We compare the performance of BayesNetRank with another Support Vector Machine(SVM) based approach called RankSVM [5]. Performance analysis is also involved in the study to identify for which kind of queries, proposed system gives results on either extremes. Evaluation results are shown using two rank based evaluation metrics, Mean Average Precision (MAP) and Normalized Discounted Cumulative Gain (NDCG).
  • Thumbnail Image
    ItemOpen Access
    Micro-level drought preparedness with information management and rural knowledge centres: a framework to support rural farm families
    (Dhirubhai Ambani Institute of Information and Communication Technology, 2010) Guntuku, Dileepkumar; Chaudhary, Sanjay; Balaji, V
    Drought and desertification are serious problems that significantly affect hundreds of millions of people and ecosystems. When drought occurs, the farm communities are usually the first to be affected because of their heavy dependence on the stored soil water. If the rainfall deficiencies continue, even people who are not directly engaged in agriculture will be affected by drought. This underscores the vulnerability of entire societies to this phenomenon; this vulnerability varies significantly from one nation to another. Although crisis management approach is routinely followed approach for providing relief, the studies on drought, carried out in different parts of the world, suggested that preparedness is better than relief and information is backbone of drought preparedness. However, the efforts have been taken for generating micro-level drought assessment and early warning is least understood until recent years. It was therefore, in this study, an attempt has been made to develop a micro-level drought preparedness framework to support rural farm families. The established practices such as Sources of Agricultural Information management (International/National/Extra-Institutional), Information and Communication Technology (ICT) Enabled Rural Knowledge Centres (RKC), Open and Distance Learning Methods, micro-level drought assessment and early Warning technique have been identified as key components in developing such framework. These components were considered as the objectives of this research study, and conducted series of studies and experiments to understand the existing approaches and needed arrangements in defining and developing proposed framework. For each finding reported in the experimental objectives, a clear chain of evidence was established supported also by interview statements. The individual micro-level drought preparedness framework components were integrated carefully, based on the series of findings, systemic analysis of the data and the continuous interpretation of the observations, to develop the proposed framework. The study concludes that the proposed framework has shown a way to improve micro-level drought preparedness by bringing various ICT tools, information management techniques, open learning approaches, and micro-level drought assessment technique under one umbrella with an intermediary entity called ICT enabled RKCs owned and run by rural farm families. The usability evaluation studies on individual components revealed that the approaches such as these will have implications in planning micro-level drought preparedness strategies. The vulnerable rural families now have the means to estimate their own vulnerability and can use the information available at ICT enabled RKCs to make more informed decisions, which offers a sounder basis for designing drought preparedness and adaptation strategies.