🎬 Auto Shorts Generator

Read this in other languages: English | KR.

An automated Python tool that takes a long-form video, identifies the most engaging highlights ("hooks"), and automatically edits them into viral short-form videos (TikTok, YouTube Shorts, Instagram Reels).

I also make CLI!!

• Interactive CLI: Easy-to-use menu-driven command-line interface.
• Smart Hook Extraction: Leverages Large Language Models (Gemini or OpenAI) to analyze transcripts and find the most engaging parts of your video.
• High-Performance Transcription: Uses faster-whisper for fast and accurate local speech-to-text conversion.
• Dynamic Face Tracking: Automatically detects and tracks faces, cropping landscape (16:9) videos into portrait (9:16) format while keeping the subject centered.
• Bouncing Subtitles: Generates TikTok-style word-by-word highlighted subtitles with a dynamic bounce effect.
• Auto Background Blur: Fills the background with a blurred version of the original video for a premium look.

1. Python 3.12 or higher.
2. FFmpeg: Must be installed and added to your system's PATH.
3. GPU (Optional but recommended): For faster Whisper transcription and NVENC video encoding (NVIDIA).

RTX 50XX Series Currently, there may be library compatibility issues with NVIDIA RTX 50-series GPUs, which can cause the Whisper transcription speed to be significantly slower than expected. This is a known issue with the underlying libraries and may be resolved in future updates.

This project uses uv for lightning-fast Python package management.

0. Install uv:

   pip install uv

1. Clone this repository:

   git clone https://github.com/liquidtii/Auto_Shorts_Generator.git
   cd Auto_Shorts_Generator

2. Install dependencies (this will automatically create a virtual environment and install packages using pyproject.toml / uv.lock):

   uv sync

Run the main script using uv to start the interactive CLI:

uv run main.py

On your first run, the tool will guide you through an initial setup. You will need to provide:

• LLM Provider: Choose between gemini or openai.
• API Key: Your Gemini API Key or OpenAI API Key (depending on your choice).
• Other Settings: Output directory, Whisper model size (tiny, base, small, medium, large), subtitle font settings, etc.

Settings are saved to settings.json so you don't have to enter them every time.

• main.py: The entry point of the application. Handles the interactive CLI, user settings, and orchestrates the video generation pipeline.
• extractor.py: Handles audio extraction from the video, speech-to-text transcription using faster-whisper, and hook analysis using the chosen LLM API.
• video_editor.py: The core video editing engine using moviepy and OpenCV. Handles face tracking, 9:16 cropping, background blurring, audio mixing, and generating word-by-word subtitle clips.

You can manually edit settings.json or update it via the CLI Menu (Option 2).

• llm_provider: "gemini" or "openai"
• model_size: Whisper model size (default: "base")
• max_duration: Maximum length of the generated short in seconds (default: 60)
• highlight_color: Color of the spoken word in subtitles (default: "yellow")

This project is open-source and available under the MIT License.