Transform your long-form content into viral-ready "Wisdom Bytes" with one click. ByteSize Sage is an AI-powered pipeline that analyzes video, audio, and text to find the most impactful segments of your content and automatically formats them for TikTok, Reels, and YouTube Shorts.
Current video tools are either manual and tedious or generic and "black-box." ByteSize Sage is different. It uses a multimodal intelligence engine to understand why a clip is valuable—whether it's the speaker's emotional intensity, the density of wisdom in their words, or the visual clarity of their face.
- Stop Guessing: Let the data rank your clips by their "Viral Potential."
- Total Control: Adjust processing weights to favor humor, wisdom, or high-energy moments.
- Privacy First: High-performance local AI (OpenAI Whisper) means your content stays on your machine.
ByteSize Sage doesn't just listen; it perceives. It scores every second of your video across four distinct dimensions:
graph LR
A[Long-Form Video] --> B(Extraction Cluster)
B --> C{Multimodal Analysis}
C -->|Audio| D[RMS Energy & Emphasis]
C -->|Text| E[Wisdom Score & Sentiment]
C -->|Vision| F[Face Detection & Framing]
C -->|Emotion| G[Intensity Tracking]
D & E & F & G --> H(Viral Scorer)
H --> I[Viral Clip Selection]
I --> J(Auto-Crop & Captions)
J --> K[9:16 Viral Byte]
- 🎯 1-Click Viral Scoring: Our custom algorithm ranks segments based on multimodal peaks.
- 📱 Smart Auto-Clipping: Automatically crops to a vertical (9:16) aspect ratio, ensuring the subject stays centered.
- 🖍️ Stylized Captions: Generates high-impact, TikTok-style overlays (Yellow/Black) out of the box.
- 📊 Real-Time Analytics: Explore the "Viral Score Timeline" and see exactly where the "heat" is in your video.
- 🔗 Hybrid Input: Paste a YouTube URL or upload raw files directly.
- Python 3.9 - 3.11
- FFmpeg (for video processing)
- ImageMagick (for captioned overlays)
-
Clone & Enter
git clone https://github.com/Akkii88/BiteSize.git cd BiteSize -
Install Dependencies
pip install -r requirements.txt
-
Launch the Engine
streamlit run app.py
Get up and running in a clean container environment:
docker build -t bitesize-sage .
docker run -p 8501:8501 bitesize-sage- Input: Paste a YouTube link or drag-and-drop your
.mp4. - Tune: Use the sidebar sliders to tell the AI what you value (e.g., more "Wisdom" vs more "Energy").
- Analyze: Let the Sage traverse your content (Sub-second processing for local snippets).
- Export: Preview the generated clips in the "Viral Clips" tab and download your favorites.
This project is built for content creators who value autonomy. We don't believe in SaaS subscriptions for something your own hardware can do.
Developed with the goal of democratizing high-quality video production through AI.