Automatically find and clip highlights from gameplay recordings
Intermediate16h1–2
FFmpegPythonTesseract.js
17/20AI Live Game Commentator
Live spoken commentary for any game based on events.
Intermediate16h1–2 people18/20 fit
Build AI Live Game Commentator for streamers and esports clubs in ~16h with 1–2 people. Stack: LLM + TTS + Tesseract.js.
Why it wins: Entertaining voice AI.
Build flow
01 · Problem
Small streams lack commentary and hype.
02 · Solution
Read game events from logs or screen text, generate exciting commentary and speak it with TTS.
03 · Build
- 1Capture events
- 2Generate lines
- 3Speak
- 4Tune timing
04 · Demo
Play a match with live AI commentary.
MVP checklist
- Event capture
- Commentary styles
- Voice output
- Highlights
Tech stack
| Tech | Role | |
|---|---|---|
| LLM | Language model | AI |
| TTS | Text-to-speech | AI |
| Tesseract.js | OCR in the browser | AI |
Reference repos
Open-source projects you can build on. Links checked 2026-10-31.
Related ideas
More GamingOne player in a werewolf-style game is secretly an AI
Intermediate16h2–3LLMSocket.IONext.js18/20Gaming
Chess Move Explainer
Explain why the engine's best move is best in plain words
Intermediate14h1Chess engineLLMNext.js17/20