← BACK TO PROJECTS
PROJECT [03]

AI Clipper

VIDEO PROCESSING / PIPELINE
TYPEVIDEO PROCESSING / PIPELINE
STATUSPRODUCTION
YEAR2026
STACKPython / Faster-Whisper / Amazon Bedrock / OpenCV

AI Clipper is an end-to-end automated pipeline designed to extract viral-worthy clips from long-form YouTube or local video content, optimized for short-form social media platforms.

The system uses Faster-Whisper for high-accuracy audio transcription, then leverages Amazon Bedrock for semantic analysis to identify the most engaging and viral-potential segments from the transcript.

A standout feature is the 'Smart Crop' capability powered by OpenCV face detection, which automatically reframes content from landscape 16:9 to portrait 9:16 format, ensuring subjects remain centered and visible for mobile-first consumption.

  • Python
  • Faster-Whisper
  • Amazon Bedrock
  • OpenCV
  • FFmpeg
ARCHITECTURE DIAGRAM
┌──────────────┐     ┌──────────────┐     ┌──────────────┐
│  VIDEO INPUT │────▶│  TRANSCRIBE  │────▶│   ANALYZE    │
└──────────────┘     │ Faster-Whisp │     │ Amazon Bedr. │
                     └──────────────┘     └──────┬───────┘
                                                 │
                                                 ▼
                                         ┌──────────────┐
                                         │  SMART CROP  │
                                         │   OpenCV     │
                                         └──────┬───────┘
                                                 │
                                    ┌────────────┼────────────┐
                                    ▼            ▼            ▼
                              ┌─────────┐ ┌──────────┐ ┌──────────┐
                              │  16:9   │ │   9:16   │ │  SHORTS  │
                              └─────────┘ └──────────┘ └──────────┘

Video Ingestion → Transcription → Semantic Analysis → Smart Crop → Export