DefinePK

Abstract

The YouTube Transcript Translation Web Application is a tool designed to extract the transcript from YouTube videos and translate it into multiple languages. This Flask-based application uses the YouTubeTranscriptApi to fetch the transcript of a given YouTube video and Deep Translator to translate the extracted text into various languages. By dividing the transcript into smaller pieces and translating each one separately, the system addresses the common problem of text in long transcripts exceeding query length limits. The field of text summarization has seen significant advancements, primarily due to progress in NLP and machine learning. Techniques range from extractive approaches, where key sentences are selected directly from the text, to abstractive methods, which generate summaries by paraphrasing the content. Tools such as BERT, GPT, and Transformer-based architectures have revolutionized summarization tasks. Previous studies have also explored video content summarization, focusing on either visual elements or transcripts. YouTube provides autogenerated transcripts for many videos, but these are often unstructured and verbose. Existing solutions for transcript summarization, such as manual editing or generic text summarizers, are time-intensive and lack context sensitivity for video-specific nuances. Current NLP-based tools may not integrate seamlessly with YouTube’s API or fail to account for timestamped content, which is critical for maintaining the structure of video narratives. The proposed YouTube Transcript Summarizer was evaluated using a dataset of transcripts from various genres, including educational videos, podcasts, and tutorials. Metrics such as ROUGE scores and user satisfaction surveys were employed to assess the system's performance. The results demonstrated an average ROUGE-1 score of 85%, indicating a high level of accuracy in retaining critical information