Browsing by Author "Jayathilake, M. V. M."

Now showing 1 - 3 of 3

Attitudes towards the Use of Web 2.0 Tools for Learning ESL: A Case Study Conducted at the Advanced Technological Institute, Gampaha.
(The Third International Conference on Linguistics in Sri Lanka, ICLSL 2017. Department of Linguistics, University of Kelaniya, Sri Lanka., 2017) Jayathilake, M. V. M.; Shantha, P. H. A. B.
With the advancement of the web technologies, almost all the fields get more benefits than previously and education is one of most benefited sectors. There is a growing interest in online classroom settings in learning second languages. There are a number of research studies that have used empirical and exploratory methods to assess and evaluate the actual impact of using Web 2.0 tools in learning English as second language, because Web 2.0 tools have shifted language learners from passive recipients into active contributors. Based on this rationale, the present study aims at exploring attitudes toward the use of Web 2.0 tools for leaning English as a second language. However, the question arises as to whether alternative higher education sector students have similar Web 2.0 experiences in ESL as their higher education sector peers. In this study, the core objective is to evaluate attitudes towards using Web 2.0 tools in Leaning English as a Second Language in the higher Education sector of Sri Lanka. Accordingly, 260 students from Advanced Technological Institute (ATI), Gampaha were selected as the sample. Framework of this study is based on Technology Acceptance Model (TAM). The study made use of a mixed method approach and the participants‟ attitudes were elicited through the use of a questionnaire which included open-ended questions to collect qualitative data and structured questions which ensured the collection of quantitative data. Responses given to open-ended questions were analyzed through coding the statements while the responses to structured questions were analyzed by calculating the frequencies. The results showed that all the respondents rated the items positively within all six dimensions of TAM. Web 2.0 tools had significant correlations to TAM dimensions and the majority of students have positive attitudes regarding the use of an interactive web environment and the learning benefits that ensued.
Comparative study of neural network based speech recognition algorithms for Sri Lankan accent
(Faculty of Science, University of Kelaniya, Sri Lanka., 2021) Jayasekara, J. K. D. R.; Jayathilake, M. V. M.; Gunasekara, S. V. S.
Speech recognition is a technology which involves processing and interpreting human speech into a written format. Advancements in technology have led to the development of sophisticated speech recognition algorithms with the use of neural networks, machine learning and artificial intelligence. Automatic Speech Recognition (ASR) is being used worldwide for developing various applications including automated devices to communicate with humans such as Alexa, Siri and Artificial Intelligence Chatbots. Several machine learning algorithms, Natural Language Processing (NLP) techniques, Hidden Markov Models (HMM) and neural networks are used to create the foremost speech recognition systems. However, most speech recognition algorithms are yet to overcome the many barriers which come along with the technology. Variations in pronunciations and accents, lack of fluency, speech clarity, speed of speech and language technicalities are just few of the challenges faced by modern day speech recognition algorithms. These problems are magnified in lesser-known languages and accents. The purpose of this research is to compare the accuracy of multiple speech recognition systems for unexplored accents such as the Sri Lankan accent. A comparison was conducted between three leading neural- network based speech recognition systems regarding their accuracy in recognizing speech spoken in a Sri Lankan accent. The primary objective of this study was to determine the system which applies the most efficient algorithm for recognizing speech with language nuances. Google Cloud Speech-to-Text, Mozilla DeepSpeech and CMU Sphinx were the three systems used in the research comparison. Quantitative secondary data was used to analyse existing speech recognition systems and their accuracy in interpreting speech in English accents. Furthermore, experimental research was conducted using primary audio data gathered using different speakers. Six selected sentences were converted to a verbal format in the form of individual audio files in the .wav format. Two versions of Sri Lankan accents were recorded for each sentence. An algorithm was designed in the Python language to calculate the Word Error Rate (WER) for each system and determine the one with the lowest error rate. Word Error Rate is a metric used to calculate the accuracy of text transcribed by speech recognition systems. The mean WERs obtained were 0.86, 1.05 and 0.59 for Mozilla DeepSpeech, CMU Sphinx and Google Cloud Speech-to-Text respectively. While the results provide conclusive evidence that the Google Cloud ASR system is the best at identifying speech in a Sri Lankan accent, it could be clearly observed that all three systems encountered difficulties when recognizing homophones and words with contradictory pronunciations. The outcome of this research indicates that although speech recognition systems have had major improvements over the years, there are still a lot more enhancements to be done in order to provide accurate and efficient speech-to-text transcriptions. The systems should be trained with larger and miscellaneous datasets which include speech from diverse languages and accents. As of now, the Google Cloud speech recognition system displays optimal performance when interpreting speech in Sri Lankan accents.
Evaluate the acceptance of web 2.0 tools for learning in alternative higher education sector in Sri Lanka.
(International Research Symposium on Pure and Applied Sciences, 2017 Faculty of Science, University of Kelaniya, Sri Lanka., 2017) Jayathilake, M. V. M.; Shantha, P. H. A. B.
Improvement of ICT has affected all the fields related to human life. All most all the fields get more benefits than previously. The education field is one of the most impacted sectors. Web 2.0 is such a technology that provides very effective web-based collaborative systems. There are number of research studies that have conducted empirical and exploratory research to assess and evaluate the actual impact of using Web 2.0 tools in learning process. Because, Web 2.0 tools have shifted language learners from passive recipients into active contributors. Based on this rationale, the present study aimed to explore the acceptance toward the use of Web 2.0 tools for learning. In this study, the core objective is to evaluate the acceptance towards using Web 2.0 tools to Learning in higher Education sector Sri Lanka. In order to do so, 280 students from Sri Lanka Institute of Advanced Technological Education was selected randomly as a sample. Framework of this study is based on Technology Acceptance Model (TAM). A conceptual model has 9 variables and associated hypotheses were designed based on the literature review and the initial primary study. The nine variables perceived usefulness, performance expectancy, social factor, behavioral intentions, prior knowledge or the use for social purpose, facilitating conditions, motivation to use, ease of use and the actual use. A questionnaire was developed from the model operationalization. The findings of the study have shown and validated what was previously found in the literature in which the majority of students is positive about the use of an interactive web environment, and found its use beneficial for their learning. Some of the validated variables perceived usefulness and prior knowledge. The conclusion of this study is the utilization of Web 2.0 facilities to stimulate participation in learning. This study will contribute to the body of knowledge on the acceptance of Web 2.0 social networking tools in learning.