Development & evaluation of different acoustic models for Malayalam continuous speech recognition

Performance of any continuous speech recognition system is dependent on the accuracy of its acoustic model. Hence, preparation of a robust and accurate acoustic model lead to satisfactory recognition performance for a speech recognizer. In acoustic modeling of phonetic unit, context information is of prime importance as the phonemes are found to vary according to the place of occurrence in a word. In this paper we compare and evaluate the effect of context dependent tied (CD tied) models, context dependent (CD) and context independent (CI) models in the perspective of continuous speech recognition of Malayalam language. The database for the speech recognition system has utterance from 21 speakers including 11 female and 10 males. Our evaluation results show that CD tied models outperforms CI models over 21%.

Phytochemical investigations on ‘black glumed’ Njavara (Oryza sativa L.), the medicinal rice, as compared to staple varieties and evaluation of their antioxidant, anti-inflammatory and anticancer effects

The study was planned to investigate the bioactive compounds in Njavara compared to staple varieties and their bioactivity to substantiate the medicinal properties. Results of the study on chemical indices, antioxidant activity and antiinflammatory activity (in vivo) of Njavara black rice bran and rice in comparison with non-medicinal varieties like Sujatha and Palakkadan Matta rice bran and rice are given. The phytochemical investigation and quantification of Njavara extracts in comparison with staple varieties are detailed in this study. The last chapter is divided in three sections (A, B and C). Section A comprises the antioxidant activity by in vitro assays like DPPH, superoxide anion radical and hydrogen peroxide scavenging activity of the compounds. Also, theoretical studies using DFT were carried out based on DPPH radical scavenging activity for understanding the radical stability and mechanism of antioxidant activity. Section B comprises the anti-inflammatory activity of the identified compounds namely tricin and two flavonolignans in both in vivo and in vitro models. Section C describes the cytotoxicity of the rare flavonolignans, tricin 4’-O-(erythro-β-guaiacylglyceryl) ether and tricin 4’-O-(threo-β-guaiacylglyceryl) ether towards multiple cancer cells belonging to colon, ovarian and breast tumours.

Speaker identification using models for phonemes

Motivation for Speaker recognition work is presented in the first part of the thesis. An exhaustive survey of past work in this field is also presented. A low cost system not including complex computation has been chosen for implementation. Towards achieving this a PC based system is designed and developed. A front end analog to digital convertor (12 bit) is built and interfaced to a PC. Software to control the ADC and to perform various analytical functions including feature vector evaluation is developed. It is shown that a fixed set of phrases incorporating evenly balanced phonemes is aptly suited for the speaker recognition work at hand. A set of phrases are chosen for recognition. Two new methods are adopted for the feature evaluation. Some new measurements involving a symmetry check method for pitch period detection and ACE‘ are used as featured. Arguments are provided to show the need for a new model for speech production. Starting from heuristic, a knowledge based (KB) speech production model is presented. In this model, a KB provides impulses to a voice producing mechanism and constant correction is applied via a feedback path. It is this correction that differs from speaker to speaker. Methods of defining measurable parameters for use as features are described. Algorithms for speaker recognition are developed and implemented. Two methods are presented. The first is based on the model postulated. Here the entropy on the utterance of a phoneme is evaluated. The transitions of voiced regions are used as speaker dependent features. The second method presented uses features found in other works, but evaluated differently. A knock—out scheme is used to provide the weightage values for the selection of features. Results of implementation are presented which show on an average of 80% recognition. It is also shown that if there are long gaps between sessions, the performance deteriorates and is speaker dependent. Cross recognition percentages are also presented and this in the worst case rises to 30% while the best case is 0%. Suggestions for further work are given in the concluding chapter.

EVALUATION OF THERMODYNAMIC PARAMETERS OF THE ATMOSPHERE BY A FORTRAN PROGRAM

Thermodynamic parameters of the atmosphere form part of the input to numerical forecasting models. Usually these parameters are evaluated from a thermodynamic diagram. Here, a technique is developed to evaluate these parameters quickly and accurately using a Fortran program. This technique is tested with four sets of randomly selected data and the results are in agreement with the results from the conventional method. This technique is superior to the conventional method in three respects: more accuracy, less computation time, and evaluation of additional parameters. The computation time for all the parameters on a PC AT 286 machine is II sec. This software, with appropriate modifications, can be used, for verifying various lines on a thermodynamic diagram

A framework for translating English text into Malayalam using statistical models

A methodology for translating text from English into the Dravidian language, Malayalam using statistical models is discussed in this paper. The translator utilizes a monolingual Malayalam corpus and a bilingual English/Malayalam corpus in the training phase and generates automatically the Malayalam translation of an unseen English sentence. Various techniques to improve the alignment model by incorporating the morphological inputs into the bilingual corpus are discussed. Removing the insignificant alignments from the sentence pairs by this approach has ensured better training results. Pre-processing techniques like suffix separation from the Malayalam corpus and stop word elimination from the bilingual corpus also proved to be effective in producing better alignments. Difficulties in translation process that arise due to the structural difference between the English Malayalam pair is resolved in the decoding phase by applying the order conversion rules. The handcrafted rules designed for the suffix separation process which can be used as a guideline in implementing suffix separation in Malayalam language are also presented in this paper. Experiments conducted on a sample corpus have generated reasonably good Malayalam translations and the results are verified with F measure, BLEU and WER evaluation metrics