Research Publications
Permanent URI for this communityhttps://rda.sliit.lk/handle/123456789/4194
This main community comprises five sub-communities, each representing the academic contribution made by SLIIT-affiliated personnel.
Browse
13 results
Search Results
Item Embargo Sinhala Speech Recognition System for Speech-Based Autism Intervention in Children Using the NAO Robot(Institute of Electrical and Electronics Engineers Inc., 2026-03-26) Bopage, H; Pulasinghe, K; Rajapaksha, SThis research focuses on the development of a Sinhala speech recognition engine tailored to identify the language content of conversations with children. The engine leverages machine learning algorithms and natural language processing (NLP) techniques to transcribe and classify speech in Sinhala. Key features include an acoustic model optimized for the nuances of Sinhala phonetics and a language model trained on datasets encompassing texts of child-directed speech. The system evaluates linguistic aspects to assess the appropriateness of content and engagement levels in child-centric dialogues. By addressing challenges such as phoneme variation and informal conversational patterns, the system aims to enhance the understanding and facilitation of Sinhala-based child interactions, promoting effective communication and developmental support.Publication Open Access The Influence Of Project Managers' Decision-Making Styles On Schedule Variance In Building Construction Projects(Faculty of Engineering, 2025-09-09) Pulasinghe, K; De Silva, PThis research focus to assess the relationship between project managers decision making styles and schedule variance in building construction projects in Sri Lanka. Timely completion of construction projects is one of the major performance indicators, yet delays are a long-standing issue in Sri Lankan construction projects. Though there are many causal factors, decision-making styles of project managers have not been studied yet, creating a significant knowledge gap with regard to their influence on project schedule variance. This study attempts to analyze the relationship between project managers' decision-making styles: directive, analytical, conceptual, and behavioural and schedule variance in Sri Lankan building construction projects. A mixed-methods research approach was adopted. Primary data were gathered by holding semi-structured interviews with nine industry practitioners and a questionnaire survey of 50 respondents covering key project roles. To analyze the data code based content analysis and descriptive statistical tools (percentage count, mean, weighted average etc.) were used and to examine the relationship between decision-making styles and schedule variance Pearson correlation was conducted. Results revealed that decision making styles play a significant role in influencing project schedules. However, it was found that directive and behavioural styles are most prevalent and successful styles in the Sri Lankan context. The data revealed both positive and negative influences of managerial decisionmaking styles on schedule performance. These results contribute to the link between leadership and project performance and make a theoretical and practical contribution by revealing decision-making as a key influence to minimize schedule variance.Item Embargo Adaptive Robotic Voice Modulation for ASD Kids: Tailored Voice Pitch, Tone, and Speed(Institute of Electrical and Electronics Engineers Inc., 2025) Panduwawala, P; Pulasinghe, K; Rajapaksha, SChildren with Autism Spectrum Disorder (ASD) often experience sensory sensitivities, particularly auditory hypersensitivity, which can make interactions and communication challenging. This study explores the customization of the NAO robot's voice pitch, tone, and speech speed using the Kaldi Speech Recognition Toolkit to align with the preferences of children with ASD. Eight distinct voice profiles were created, offering a range of variations in pitch, tone, and speech speed. Parents or caretakers were asked to select the voice profile they felt would be most suitable for their child. Based on this feedback, we created a spectrum of voices tailored to each child's needs. Results indicate that medium-pitch and moderate-speed combinations are most effective in enhancing engagement, with Voice 2 emerging as the preferred profile. The findings underscore the potential of adaptive voice modulation in improving robotic interactions for ASD therapy and highlight opportunities for further research in real-time adaptability and long-term impact assessment.Item Open Access Designing Culturally Adaptive Emotional Gestures to Enhance Child-Robot Interaction with NAO Robots in ASD Therapy(Institute of Electrical and Electronics Engineers Inc., 2025) Manukalpa, C.S; Pulasinghe, K; Rajapakshe, SIntegrating social robots into human-robot interactions demands advancements in natural language processing, navigation, computer vision, and expressive gestures to foster meaningful interactions. However, a gap remains in designing culturally relevant and developmentally appropriate gestures, particularly in the Sri Lankan context. Autism Spectrum Disorder (ASD), a neurodevelopmental condition impacting early education, often remains underdiagnosed, exacerbating learning challenges. This study introduces a novel approach utilizing robot-child interactions for ASD screening to minimize such delays. Expressive gestures were developed for the NAO6 humanoid robot to engage Sinhala-speaking children aged 2 to 6 years, including those with ASD, in Sri Lanka. Using the NAOqi Python API and Choregraphe simulator, culturally aligned gestures expressing emotions like happiness, sadness, fear, anger, and more were designed and synchronized with voice and LED effects. Pilot studies with typical children demonstrated the significance of linguistic and cultural alignment in enhancing engagement, emotional response, and trust. By addressing cultural nuances and advancing early ASD screening, this framework holds potential for broader applications in education, therapy, and diagnosis, improving human-robot interactions globally.Item Embargo Child's Age Range Prediction Using Sinhala Speech Recognition System(Institute of Electrical and Electronics Engineers Inc., 2025) Kathriarachchi, A; Pulasinghe, KThis study predicts the age range of a child speaking Sinhala by analyzing voice characteristics and acoustic features. Identifying speech impairments in children aged 6 to 72 months is critical for early intervention, mainly when using a system that recognizes their native language. The developed system generates accurate insights to assist speech pathologists in diagnosing speech disorders. A Multilayer Perceptron neural network is proposed for age group prediction, leveraging Mel Frequency Cepstral Coefficients (MFCC) and pitch features to enhance recognition accuracy. The system demonstrated an overall accuracy rate of 77% in age range identification, providing a valuable tool for healthcare professionals to evaluate and monitor speech development in Sinhala-speaking childrenItem Embargo Advancing Speech Therapy for Sinhala-Speaking Children with Autism Spectrum Disorder Through an Intelligent Dialog System(Institute of Electrical and Electronics Engineers Inc., 2025) Jayawardena, A; Pulasinghe, K; Rajapakshe, SThis paper presents a dialog system integrated with a NAO socially assistive robot, designed to support Sinhala-speaking children with Autism Spectrum Disorder (ASD). The system leverages a pipeline-based architecture implemented using the RASA framework, consisting of Natural Language Understanding (NLU), Dialog Management (DMU), and Natural Language Generation (NLG) units. The NLU unit processes user input by identifying intents, entities, and dialogue acts, incorporating custom tools like the SpokenSinhalaVerbTokenizer for handling spoken Sinhala. The DMU includes a Dialog State Tracker (DST) to maintain conversation context and a Dialog Policy Generator, which employs rule-based, TED, and UnexpecTED policies to adapt conversation flows dynamically. The NLG unit generates natural responses to foster interactive and goal-oriented conversations. Integrated with the NAO robot, the system engages children through meaningful dialogues, such as discussing toy preferences, aiming to enhance social interaction and communication skills. This work highlights the potential of conversational AI and robotics in therapeutic interventions for ASD in low-resource languages.Publication Embargo Step-by-Step Process of Building Voices for Under Resourced Languages using MARY TTS Platform(IEEE, 2022-12-09) Senarathna, M; Pulasinghe, K; Reyal, SThis paper presents a comprehensive guide for creating synthetic voices to support under resourced languages for the MaryTTS platform. Although researchers have extensively contributed in the domain of speech synthesis, the lack of a thorough documentation hinders the voice building process for languages not yet supported by MaryTTS, complicating the implementation process for users with inadequate knowledge in the field of Text-to-Speech (TTS). The step-by-step process discussed in this study is further demonstrated with the creation of a synthetic voice for the Sinhala language, with unit selection as the voice building approach. A Sinhalese voice was generated with an intelligibility score of 91.7% upon evaluation with Diagnostic Rhyme Test (DRT). Comparison with ground truth data proved a close approximation to human speech where the intelligibility score was identified as 97.9%, when tested with the same participants. The Mean Opinion Score (MOS) revealed a naturalness level of 2.993, indicating a moderately high speech quality for the proposed system in comparison with the ideal score of 4.972.Publication Embargo Sinhala Conversational Interface for Appointment Management and Medical Advice(IEEE, 2020-12-10) Rajapakshe, D. D. S; Kudawithana, K. N. B; Uswatte, U. L. N. P; Nishshanka, N. A. B. D; Piyawardana, A. V. S; Pulasinghe, KThis paper proposes an intelligent conversational user interface to assist Sinhala speaking users to make appointments with doctors and to obtain medical advices. This Sinhala Conversational Interface for Appointment Management and Medical Advice (SCI-AMMA) consists of Speech Recognition unit, Query Processing unit, Dialog Management unit, Voice Synthesizer unit, and User Information Management unit to handle user requests and maintain a meaningful dialogue. The SCI-AMMA gets the users' speech utterances and recognize the language content of it for further processing. Language content is further processed using query processing unit to identify users' intent. To fulfil the users' intent, a reply is generated from Dialogue Management Unit. This reply/answer will be delivered to the user by means of a voice synthesizer. The proposed system is successfully implemented using state of the art technology stack including Flutter, Python, Protégé and Firebase. Performance of the system is demonstrated using several sample scenarios/dialogues.Publication Embargo Vehicle Insurance Policy Document Summarizer, AI Insurance Agent and On-The-Spot Claimer(IEEE, 2021-04-02) Samarasinghe, H. T. D; Herath, N. A. D. M; Dabare, H. S. S; Gamaarachchi, Y. R; Pulasinghe, K; Yapa, PThis paper proposes an automated vehicle insurance policy summarizing application. “Explain to Me” is one such software/tool which enable you to summarize the content of documents regarding vehicle insurance policies by using the NLP, machine learning and deep learning applications. The program targets mainly insurance users and suppliers of insurance services. Due to the increase of vehicle accidents, the vehicle insurance industry has gained more popularity currently. Therefore, different insurance companies have introduced a variety of insurance policies to customers. Vehicle insurance policy documents consist lot of insurance terms that should be read with more attention. As the main objective, this system filters unnecessary data in the particular document, and finalize a summary as the output. As another major component, the application “On the spot claimer” which is never before in Sri Lankan vehicle insurance industry, is another major part of this project that works as suggesting the most relevant insurance claiming that can be claimed by the user after detection of the type of damage through mobile phone camera. Another part of this research project, the function known as the Recommender, which works along with the summarization tool, is a recommendation system with a view of recommending more favorable rules for the assertion of alternatives that exist in the corresponding, equivalent documents of other companies. Finally, in order to interact with custody concerns about how to insure an automobile, CNN, which are based on the extraction of images, are used for the implementation of the ETM system in NLP.Publication Embargo Automated Sinhala Speech Emotions Analysis Tool for Autism Children(IEEE, 2021-08-11) Welarathna, K. T; Kulasekara, V; Pulasinghe, K; Piyawardana, V— Autism Spectrum Disorder (ASD) is a neurological disorder that impairs children's development and symptoms that can be noticed in early childhood. One of the main diagnosis characteristics of ASD is the child having unusual emotions and expressions during social interactions. The main problem is how to distinguish these symptoms. Only 14 out of 100 Autistic kids, before they reach the age of 24 months, get medical treatments since the unavailability of resources to identify them early. If they can be recognized early, a therapeutic engagement can be done to help them overcome those issues in social interactions, when they reach school-going age. The focus of this research is to develop a tool to screen atypical children from typical children. This research attempts to recognize the correct emotion of a child, while the child is talking. The input audio stream of children was normalized into a specific range, sub-framed into 2s length for language-independent, noise reduction, and age independence features, and extracting the most effective 40 audio features. The Convolutional Neural Network (CNN) based model classifies eight different emotions of sad, disgust, surprise, neutral, happy, calm, fear, and angry with an accuracy matrix of F1 score of 0.90, even in the uncontrol environment. If the classifying emotions have small frequency variances, the trained model has the ability to handle them.