Backend development and deployment of local-based speech recognition system

Abstract

This Final Year Project focuses on adapting and deploying the backend architecture of an existing full-stack web application, initially designed for online use, to a local offline platform. The backend system facilitates functions such as offline audio transcription, live transcription via microphone input, and managing past transcriptions by leveraging NTU Speech Lab’s speech recognition application programming interfaces. New admin dash- board functionalities were also developed, enabling administrators to efficiently modify user permissions and manage features.

The primary objective of this transition to a local deployment is to enhance data privacy, minimize internet dependency, and optimize system performance. The development process involved re-configuring the backend architecture, optimizing local resource management, and ensuring seamless integration with NTU Speech Lab’s speech recognition application programming interfaces in an offline environment.

This report details the stages of the project, including requirements analysis, system redesign, local environment setup, backend implementation, testing, and deployment, with a focus on maintaining the original user experience while achieving local execution.