Inspiration
Content creators spend hours turning one long-form video into multiple Shorts, Reels, and TikToks. We built COOK to automate this repetitive process and help creators turn one video into a week of content.
What it does
COOK transforms a long-form video into:
- Short-form 9:16 clips
- Real timestamped captions
- High-potential moments
- AI-generated hooks
- Platform-ready content
- A weekly content calendar
- Downloadable content packages
How we built it
COOK uses Next.js, React, TypeScript, Tailwind CSS, FastAPI, Python, FFmpeg, Supabase, and Groq Whisper. Groq Whisper provides real speech transcription with timestamps, while our processing pipeline identifies key moments, creates vertical clips, and generates synchronized captions and hooks.
Challenges we ran into
The biggest challenge was reliable, accurate transcription and caption synchronization. Local speech-to-text processing caused performance issues, so we integrated Groq Whisper and built timestamp-aware ASS subtitles to keep captions synchronized with the original speech.
Accomplishments that we're proud of
We built an end-to-end workflow where a single video can become multiple creator-ready assets without manually editing each one. COOK combines transcription, clip selection, captions, hooks, and content planning in one platform.
What we learned
We learned that building AI products is not just about choosing a model. Reliable pipelines, accurate timestamps, error handling, media processing, and connecting multiple stages together are equally important.
What's next for COOK
We plan to add dynamic face tracking, automatic B-roll selection, smarter retention analysis, and one-click publishing to YouTube Shorts, Instagram Reels, and TikTok.
Built With
- fastapi
- ffmpeg
- groq
- next.js
- postgresql
- python
- react
- short-form-video
- speech-to-text
- supabase
- tailwind-css
- typescript
- video-editing
- whisper
Log in or sign up for Devpost to join the conversation.