gnanaprakash2918/ai-media-indexer ? reverse-engineered prompt

Build me a media library search app that can index a folder or uploaded videos, audio, and images, then let me search it in plain English. I want to find things like a person in a red shirt, someone speaking near a door, cheering in the background, or text that appears on screen, and get back the best matching clips with timestamps and previews.

Please make it process files in the background, extract speech, detect scenes, read visible text, recognize faces, and keep the results searchable in a fast way. I’d like a simple web app for browsing, searching, and opening results, plus a backend that handles the indexing and retrieval. If it helps, look up current docs online for any tricky parts and build it in the most practical way.

Please also include a clean setup using Docker so I can run everything locally without too much hassle.