RAG-TranSum a RAG based text summarizer and translator for Audios from vid clips

Working

The application processes audio and video files to extract transcribed text, saving it locally. Text data is not converted to vector embeddings or stored in a vector database. Users have the option to summarize or translate the extracted text, which is processed by the model to generate the output.

Tasks

The project involved the following subtasks:

• Task 1: Extracting text from audio and video files.
• Task 2: Creating a vector DB for storing embeddings(chromaDB).
• Task 3: Developing a summarizer and translator using open-source Large Language Models (LLMs).

RAG pipeline visualized