MEETTALK: VIRTUAL MEETING APPLICATION WITH LONG SHORT-TERM MEMORY RECURRENT NEURAL NETWORK ALGORITHM

Abstract

This thesis presents a virtual meeting application that implements Long Short-Term Memory Recurrent Neural Network for captioning and integrates with Agora SDK to enable Real-time Audio, Video, and Messaging. The purpose of this project is to create a virtual meeting application that contains all necessary features for meeting use while solving one major problem that has been overlooked by popular video meeting applications. The one particular problem that was referred to previously is those applications do not provide a separate section for private chat or even do not have the feature appearing in the application. The private chat feature in Zoom is designed awkwardly and in an unconventional way thus making Zoom very prone to human error. Plenty of people are using virtual meeting applications on a daily basis and by increasing the user experience will indirectly give a lot of benefit to the society.