Overview
A short-form video production CLI that automates everything from video generation to YouTube upload for marketing an AI fortune-telling app. Generates videos with Veo 3.1, extracts subtitles with Whisper, then applies LLM correction, face-detection cropping, BGM mixing, and uploads.
Tech Stack
- Video Generation: Google Veo 3.1
- Speech Recognition: MLX Whisper
- Video Post-processing: FFmpeg, MoviePy, OpenCV
- Artifact Removal: EasyOCR + LaMa Inpainting
- Audio Separation: Demucs
- Upload: YouTube API (OAuth2)
- Testing: pytest (397 tests, 91% coverage)