I want to convert pdf/docx/images to text so i can ask gpt a question that is related to this file. How would I do that efficiently?