Design And Development Of An Ai-Based Automatic Captioning System For Web Video Advertisements To Improve Accessibility For Deaf Users

Abstract

The lack of subtitles in online video advertisements limits access to audiovisual information for deaf and hard-of-hearing users. Manual subtitle creation is often time-consuming, costly, and requires specialized software, making it impractical for many content creators. This study develops a web-based automatic captioning system by integrating the OpenAI Whisper Automatic Speech Recognition (ASR) model to automatically generate synchronized subtitles from online video advertisements. The proposed system enables users to generate subtitles by simply providing a video URL and selecting the transcription language through a browser-based interface. The backend automatically retrieves the video, extracts its audio, transcribes speech using the Whisper API, and converts the transcription into a SubRip Subtitle (SRT) file with synchronized timestamps. The developed application was evaluated using User Acceptance Testing (UAT) and Black-box Testing to verify its functional performance. The evaluation results showed that all core functionalities, including video URL input, language selection, subtitle generation, subtitle display, and subtitle download, successfully passed the testing process and operated according to the expected specifications. The developed application was evaluated using UAT and Black-box Testing. All core functions, including URL input, language selection, subtitle generation, display, and download, operated as expected, demonstrating the system’s functional feasibility. The system also has potential to support digital accessibility for deaf and hard-of-hearing users, although direct user evaluation and quantitative ASR accuracy assessment were beyond the current scope.