jipraks/yt-short-clipper

What it solves

YT Short Clipper v2 is a Windows desktop application designed to automate the process of converting long-form YouTube videos into short-form, 9:16 portrait clips. It eliminates the manual effort of finding highlights, reframing video for mobile screens, and adding captions.

How it works

The app uses a combination of tools to process videos:

1. Highlight Detection: It fetches a video's subtitle transcript via yt-dlp and sends it to an OpenAI-compatible LLM to identify the most engaging segments (typically 58–120 seconds).
2. Visual Reframing: It converts the video to a 9:16 aspect ratio using three modes: face tracking via MediaPipe, centered with black bars, or centered with a blurred fill.
3. Captioning: It burns word-by-word captions directly from the original YouTube subtitle track, avoiding the need for separate transcription APIs.
4. Integration: Users can provide "AI direction" (specific time ranges or topics) to steer the LLM's selection process.

Who it’s for

Content creators and social media managers who want to quickly turn their long-form YouTube content into shorts, reels, or TikToks on Windows 10/11.

Highlights

• Flexible AI Providers: Supports any OpenAI-compatible endpoint, including OpenAI, Google Gemini, Groq, and local options like vLLM or Ollama.
• AI-Driven Steering: Allows users to specify exact time ranges or topics to force the LLM to pick specific clips.
• Face Tracking: Uses MediaPipe to ensure the speaker remains centered in the portrait crop.
• No External Transcription: Leverages existing YouTube subtitles for fast, word-by-word captioning.
• Direct Upload: Optional integration with the Repliz platform for scheduling and uploading clips.