Creating Narrated AI Demo Videos for Hackathon Projects

The author describes a repeatable workflow for producing narrated demo videos for hackathon projects without relying on a personal recording. A locally served language model running on vLLM helps turn rough project notes into a structured, scene-by-scene script with a hook, problem, solution, demo, technology overview, and closing. The approved narration is then converted into speech using the edge-tts Python package, which interfaces with Microsoft Edge's text-to-speech service. The actual product walkthrough is captured through screen recordings so the demo stays authentic, and the audio, footage, and supporting assets are assembled in a video editor. The author notes that local inference is not automatically free and that edge-tts is unofficial, so the process is meant to be reusable across different projects.