NaufalRizqullah /opensource-clipping
🎬 Ultimate AI Auto-Clipper — Transform long-form videos & podcasts into viral shorts! Features smart face-tracking, kinetic karaoke subtitles, contextual B-roll, AI voice-overs, and auto-uploaders (YouTube/FB). Powered by Whisper, Gemini, MediaPipe, & Pyannote.
★ 299⑂ 135👁 2⑃ 5● 5▤ 0
299
stars
135
forks
2
watchers
5
prs
5
issues
0
releases
README
OpenSource Clipping
Ultimate AI Auto-Clipper & Teaser Generator
An open-source content factory that transforms long-form videos into cinematic short-form highlights with hook teasers, karaoke subtitles, and auto-thumbnails.
🇮🇩 Baca dalam Bahasa Indonesia
·
Report Bug
·
Request Feature
---
🟢 Actively Maintained: contributions, bug reports, and feature PRs are welcome.
---
✨ Features
| Feature | Description |
|---|---|
| AI Transcriber | Word-level transcription using Faster-Whisper (large-v3) |
| AI Content Curator | Google Gemini analyzes context, picks the most viral moments, and generates metadata |
| Smart Auto-Framing | Face-tracking via MediaPipe BlazeFace (Short-Range) with Smooth Pan, Deadzone & anti-jitter algorithms |
| Cinematic Teaser Hook | 3-second hook with dark overlay, cinematic bars, and TV Glitch transition |
| Karaoke Subtitles | Word-by-word highlighted .ASS subtitles (Alex Hormozi / Veed style) |
| Kinetic Typography | AI-driven word emphasis with bounce/stagger animations & dual-font system |
| B-Roll Integration | Auto-fetches contextual stock footage from Pexels with crossfade & Ken Burns (Supports Hybrid, Split-Screen & Camera-Switch) |
| Multi-Hook Intro (V2) | Creates high-retention 3-4 micro-hook intros with flash/glitch transitions |
| Smart Segment Trimming | AI dynamically selects the best segments to cut out boring/silent parts |
| Auto-BGM & Ducking | Local BGM asset pool (assets/bgm/) with 2 modes: sidechain ducking (BGM auto-lowers during speech) or background (constant low volume). MP3 files auto-loop if shorter than the video |
| Auto-Thumbnail | Frame extraction with dark overlay and large title text |
| Watermark Engine | Text & image watermarks with adjustable position (9 anchors), padding, opacity, and auto-scaling |
| Cross-Platform Metadata | YouTube title/description/tags + TikTok caption — all in English |
| Auto YouTube Uploader | Automatically upload highlight clips to YouTube with scheduling support and full metadata (optional) |
| Auto Facebook Reels Uploader | Upload Reels to Facebook Pages via Meta Graph API with smart scheduling — first clip publishes immediately, subsequent clips auto-schedule at configurable intervals (optional) |
| Podcast Split-Screen | Auto speaker diarization via Pyannote with top-bottom split-screen layout for podcasts (9:16). Supports 3+ speakers across multiple scenes with per-speaker frozen frame fallback |
| Podcast Camera Switch | Auto active-speaker detection with scene-aware switching — full 9:16 crop focuses on whoever is talking; blurred pillarbox only when speakers in the same scene talk simultaneously (9:16) |
| AI Voice-Over | Converts auto-clips into original commentary/reaction videos using Gemini (script generation) and edge-tts (free text-to-speech), complete with audio ducking, text override, and ambient edge glow |
🎬 NEW: Story Clip Mode (--story-mode)
Need to assemble a narrative fro