Automatic Video Subtitle Generation System through AI

The rapid growth of digital video content on online platforms has created a significant demand for automated subtitle generation systems to improve accessibility, content understanding, and multilingual communication. Subtitles play an important role in helping hearing-impaired individuals, non-native speakers, and viewers in noisy environments understand video content effectively. However, manual subtitle creation is time-consuming, labor-intensive, and costly, especially for multilingual videos. To address this problem, this project presents an AI-powered multilingual subtitle generation system that automatically generates and embeds subtitles for video content in Telugu and English. The proposed system uses speech recognition and natural language processing technologies to convert spoken audio from video into text subtitles. The system integrates the speech recognition model OpenAI Whisper for automatic speech transcription and translation, and the multimedia processing tool FFmpeg for audio extraction and subtitle embedding. The entire system is deployed as a web application using Flask, allowing users to upload videos and download subtitled videos through a browser interface. The system workflow begins with video upload through the web interface. The uploaded video is processed using FFmpeg to extract the audio stream in a 16 kHz mono PCM format suitable for speech recognition processing. The extracted audio is then converted into log-mel spectrogram features and processed using a transformer-based encoder-decoder architecture of the Whisper model. The model generates time-stamped text segments directly from speech audio, eliminating the need for separate alignment tools.