{"slug": "how-i-create-ai-demo-videos-for-my-hackathon-projects", "title": "How I Create AI Demo Videos for My Hackathon Projects", "summary": "A developer documented a repeatable workflow for producing narrated hackathon demo videos by running a local language model through vLLM's OpenAI-compatible API to draft scene-by-scene scripts, then generating voiceover with the edge-tts Python package against Microsoft Edge's online neural voices. The pipeline keeps screen recordings of the real application authentic and assembles narration, footage and supporting assets in a video editor, with the developer noting that local inference still requires suitable hardware and that edge-tts is an unofficial interface dependent on an internet connection.", "body_md": "*Using a local LLM with vLLM, Microsoft Edge TTS, and AI tools to turn project demos into clear, engaging videos.*\n\nBuilding a hackathon project is only half the work. You also need to\n\nexplain the problem, show the product working, and make the value\n\nobvious to judges.\n\nI wanted a repeatable way to create narrated demo videos without\n\ndepending entirely on expensive cloud-based LLM APIs or recording every\n\nvoiceover myself.\n\nMy workflow combines:\n\nThe goal is not to make an AI-generated video pretend to be the\n\nproduct. The goal is to explain the real product clearly, using AI to\n\nspeed up production.\n\nI run a language model locally and serve it through **vLLM**. I can send\n\nit the project description, feature list, technical stack, and key user\n\nflows.\n\nI use the model to:\n\nvLLM provides an OpenAI-compatible API, which makes it convenient to\n\nconnect to scripts and applications that already support that API\n\nformat.\n\nDocumentation: [vLLM documentation](https://docs.vllm.ai/)\n\n**A practical note:** local inference is not automatically free. You\n\nstill need suitable hardware, and performance depends on the model,\n\nquantization, context length, and available memory.\n\nAfter the script is ready, I pass the narration text to **Microsoft Edge TTS** to generate speech.\n\nThe Python package [`edge-tts`](https://github.com/rany2/edge-tts) makes\n\nit straightforward to generate audio with available Microsoft neural\n\nvoices.\n\nWhy I use it:\n\n**Important:** `edge-tts` is an unofficial interface to Microsoft's\n\nonline text-to-speech service. It is not a fully local TTS engine, so\n\nthe narration step requires an internet connection and is subject to the\n\nservice's terms and availability.\n\nI can use other AI tools where they genuinely help---for example, to\n\ncreate a visual concept, simplify a diagram, suggest scene transitions,\n\nor generate subtitles.\n\nI keep the actual product walkthrough authentic: screenshots and screen\n\nrecordings should show the real application, not invented UI.\n\nI start with the material I already have:\n\nThe more concrete the input, the less generic the script will be.\n\nI ask the local LLM to structure the video into a few clear sections:\n\nI review the script before generating audio. AI-generated claims should\n\nbe checked against what the project actually does.\n\nI send the approved narration to Edge TTS and save the resulting audio\n\nfile.\n\nI keep the script divided into scenes rather than generating one huge\n\naudio file. This makes it easier to replace a sentence, adjust pacing,\n\nor rearrange the video later.\n\nI record the application running and capture the most important\n\nworkflows.\n\nA useful demo usually shows:\n\nI avoid long loading screens, unnecessary cursor movement, and repeated\n\nactions.\n\nI combine the screen recording, narration, and supporting assets in a\n\nvideo editor. I add captions, labels, transitions, and background music\n\nonly where they improve clarity.\n\nI also check that the voiceover matches the action on screen. A polished\n\nvideo is not useful if the narration describes something the viewer\n\ncannot see.\n\nThe following diagram shows the content-production flow. The project\n\nrecording and AI-generated assets join the narration during editing.\n\n``` php\nflowchart TD\n    A[Project notes, features, and tech stack] --> B[Local LLM served by vLLM]\n    B --> C[Scene-by-scene demo script]\n    C --> D[Review and edit script]\n    D --> E[Microsoft Edge TTS]\n    E --> F[Voiceover audio]\n\n    G[Real application] --> H[Screen recording]\n    I[Other AI tools] --> J[Supporting visuals and assets]\n\n    F --> K[Video editor]\n    H --> K\n    J --> K\n    K --> L[Final hackathon demo video]\n```\n\nThere are trade-offs. vLLM needs compatible hardware and setup. Edge TTS\n\nrelies on an online service. AI visuals can also take time to review. I\n\nuse each tool where it saves time rather than adding complexity for its\n\nown sake.\n\nThere is room to take this workflow further:\n\nThe long-term idea is a pipeline where project information and a screen\n\nrecording become the starting point for a nearly finished demo\n\nvideo---with a human reviewing the story and final edit.\n\nA hackathon project deserves a demo that makes its value easy to\n\nunderstand. My combination of a local LLM served with vLLM, Microsoft\n\nEdge TTS, and other AI tools helps me create narrated project videos in\n\na repeatable way.\n\nThe important part is the balance: use AI to speed up scripting and\n\nproduction, but show the real application and keep the claims honest.\n\n**Build the project. Tell the story. Show the proof.**", "url": "https://wpnews.pro/news/how-i-create-ai-demo-videos-for-my-hackathon-projects", "canonical_source": "https://dev.to/sumit0rn/how-i-create-ai-demo-videos-for-my-hackathon-projects-33l0", "published_at": "2026-11-14 12:39:17+00:00", "updated_at": "2026-11-14 12:51:01.632557+00:00", "lang": "en", "topics": ["ai-tools", "large-language-models", "ai-products"], "entities": ["vLLM", "Microsoft Edge TTS", "edge-tts", "Microsoft"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/how-i-create-ai-demo-videos-for-my-hackathon-projects", "markdown": "https://wpnews.pro/news/how-i-create-ai-demo-videos-for-my-hackathon-projects.md", "text": "https://wpnews.pro/news/how-i-create-ai-demo-videos-for-my-hackathon-projects.txt", "jsonld": "https://wpnews.pro/news/how-i-create-ai-demo-videos-for-my-hackathon-projects.jsonld"}}