Kaung Myat Kyaw (nickname: KM) is an undergraduate student studying
Computer Science at King Mongkut's University of Technology Thonburi. He
was an summer intern at
the Innovative Cognitive Computing (IC2) Research Center. He is involved in multiple research and integration projects related
to Large Language Models and Computer Vision. He is currently seeking
collaboration opportunities and research internships.
Contact - kaungmyat.kyaw@kmutt.ac.th
BudgetTracker is a Java-based command-line tool designed to help users manage their daily and monthly expenses. It allows you to set a daily budget, track spending across different categories, and export your data to CSV for easy analysis. The tool is built using Object-Oriented Programming (OOP) principles and provides a simple yet effective way to stay on top of personal finances.
A Flutter application that automatically generates multiple-choice quizzes for YouTube videos. It uses the YouTubeTranscript API, integrates Llama2 7B through Ollama, and is deployed using Flask on a server.
A simulator website where users can backtest their trading algorithms. It is implemented using backtrader, Flask and mysql database.
A robot that spreads climate change awareness by initiating conversations with detected persons using GPT-4, Google TTS, and STT models. It is implemented on Raspberry Pi 4 and uses the EfficientDet-Lite0 model for human and gesture recognition.
A portable embedded system for tracking items stored in refrigerators and estimating their expiry dates using a convolutional neural network inspired by VGG Net architecture.
A system that automatically downloads computer vision-related research papers and analyzes the datasets used. It uses Selenium, LangChain, and GPT-3.5 Turbo for automation and data extraction.
A Kaggle competition to solve puzzles (Rubik's Cube, Wreath, Globe) using the Iterative Deepening A* (IDA*) algorithm, achieving a top 5% ranking and a silver medal.
A project to classify brain activities from EEG and Spectrogram data using EfficientNetB0, with regularization and data augmentation techniques to improve robustness.
A system for managing university student data, allowing for the addition, deletion, and editing of courses and student records, developed using SQL, React JS, and Express JS.
An application for assigning tasks and rewarding employees based on performance, deployed using AWS Microservices. It won the first runner-up at the AWS Build on Myanmar 2020 Hackathon.
In this paper, we introduce ConversaSynth, a framework designed to generate synthetic conversation audio using large language models (LLMs) with multiple persona settings. The framework first creates diverse and coherent text-based dialogues across various topics, which are then converted into audio using text-to-speech (TTS) systems. Our experiments demonstrate that ConversaSynth effectively generates high-quality synthetic audio datasets, which can significantly enhance the training and evaluation of models for audio tagging, audio classification, and multi-speaker speech recognition. The results indicate that the synthetic datasets generated by ConversaSynth exhibit substantial diversity and realism, making them suitable for developing robust, adaptable audio-based AI systems.
(Link to Project Page)