We need to create a pipeline to generate transcripts from gemini-3-flash model from google. Now we would need a proper prompt which i've crafted here ed here