http://dlib.iit.ac.lk/xmlui/handle/123456789/1037 2026-04-28T14:25:29Z http://dlib.iit.ac.lk/xmlui/handle/123456789/1060 Multi - accent speech recognition for Tamil English mixed language system Mathusagar, Ragavan " Speech recognition has been a hot topic as the intelligent era has developed. Even though numerous automated speech recognition (ASR) programs have been available, a significant number of them do not support Tamil with full features. And when it comes to practical usability, just supporting only Tamil language words will not be an option since the English words are often used in the day-to-day conversation.But still considering a mixed language scenario will not totally satisfy the scenario without considering the different accent in language. So that the research has been done by considering both Tamil-English mixed language and accented speech. Data has been collected from two group of people with different Tamil accent (Sri Lankan Tamil , Indian Tamil) and the ASR models have been created using the open sourcetool “Kaldi”.“Mel Frequency Cepstral Coefficient”(MFCC) has been used for feature extracting and Hidden Markov Model (HMM) and Gaussian Mixture Model (GMM) based monophone and triphone models were created and tested. From the triphone model, the best model has been selected and used for the hybrid model creation by replacing the GMM with neural network. Model accuracy has been compared based on the WER and SER value for each model and also benchmarked with the previous systems. The results showedaccuracy improvement for the hybrid modelcompared to triphone models. Key words :Automatic speech recognition, Neural network , Acoustic models, Kaldi, Mixed language ASR" 2021-01-01T00:00:00Z http://dlib.iit.ac.lk/xmlui/handle/123456789/1059 HR Predictor Employee Turnover Prediction System Goutheman, Jayasai " In this connected era, High amount of employee turnover has been one of the major problem that is faced by HR professionals of companies. Employees are critical resources of the organizations, and retaining them has been an critical success factor for the companies’ success and continues growth. High turnover could be an indication that employees are unsatisfactory with the workspace. To compete in the workforce, companies are continuously working hard to recognize them self as employee friendly company to attract more employees. Employee’s satisfaction is one of the main indicator to identify the organization value. The primary requirement to predict the employee turnover is data. The organization have started heavily invest in the data management and information systems, including HR Information Systems (HRIS). They contain massive amount of data related to employees. Utilizing data visualization techniques and machine learning techniques, the management can figure out implicit patterns to predict turnover. This study used combination of approaches to build an industry level prediction system such as: Hyper parameter optimization, imbalance correction and hybrid model. As the result of experimenting 7 different types of machine learning algorithm, the hybrid model was composed of Random Forest and Extreme Gradient Boosting Algorithms. The hybrid model proposed has been able to achieve accuracy of 97.83% outperforming the single models. " 2021-01-01T00:00:00Z http://dlib.iit.ac.lk/xmlui/handle/123456789/1058 Improve horizontal pod auto scaling in container orchestration to adopt frequent oscillation in service requests Premanantha, Divya " Containers are lightweight stand-alone self-contained units that package their dependencies together for fine-grained resource sharing. It is a platform as a service that uses OS-level virtualization. Due to this managing container lifecycle by automating scaling, scheduling and managing the health of the container has become critical tasks of a container as a service platform such as Kubernetes container orchestrater. A flexible infrastructure is required for applications with dynamic workloads to leverage performance measures and minimize resource costs. Request with various fluctuation patterns has an impact on autoscaling the deployment to cater for the resource requirements. This project finds a solution to adopt frequent service oscillation during service requests into consideration when performing horizontal pod autoscaling in Kubernetes. " 2021-01-01T00:00:00Z http://dlib.iit.ac.lk/xmlui/handle/123456789/1057 Measuring customer satisfaction using vocal emotion recognition for Sinhala language Ranathunge, Thisari Savidya " In a competitive market world, organizations try to form good customer relationships to improve their sales by improving customer satisfaction. Most of the conventional methods to measure customer satisfaction are a hassle for the customer and do not always generate accurate results. With the rapid development of technology, a new way to measure satisfaction is Speech Emotion Recognition (SER) approach. Currently, it is both a challenging and emerging area. Even though it is a growing area, there’s no research about speech emotion recognition in the Sinhala language. Another issue is the lack of standard speech emotion data corpus for this area in the Sinhala language. To overcome this limitation, this study proposes an approach to detect emotions embedded in Sinhala language speech and then the detected emotions are then mapped to a satisfaction status. The main emotion classes used in this study are happiness, anger, sadness, and emotionless state plus considering speaker of the gender. The speech data corpus is formed using Sinhala wide screen movie speech data. This study uses CNN (1D) classification to perform the emotion classification task. The analysis results show that the model’s overall accuracy to identify the emotion along with gender is 62.58% and it resulted in 56% of precision, 63% of recall, and 57% of F1 score." 2021-01-01T00:00:00Z