BiliNote: Convert YouTube and Bilibili Videos Into Markdown Notes

BiliNote is an open-source, AI-powered note-taker designed specifically for video content. By providing a link—whether from Bilibili, YouTube, TikTok, or a local file—the tool generates a clean, structured Markdown summary. Automated screenshots and timestamped jump links are integrated by default to streamline your workflow.

Platform Agnostic. Full support for Bilibili, YouTube, TikTok, and local video files, with more sources in development.

Customizable Formatting. Take full control over how your notes are structured and presented.

Multimodal Analysis. The AI interprets visual data alongside audio to provide a more accurate summary.

Version History. Maintain and track previous drafts of your notes for every video processed.

Flexible Model Integration. Configure BiliNote to work with any GPT-compatible API endpoint.

On-Device Transcription. Runs Fast-Whisper locally on your own hardware to ensure privacy and reduce costs.

Intelligent Summaries. The underlying model distills video content into concise Markdown. It automatically embeds screenshots and creates direct links to specific moments in the source video.

Task Management. Easily revisit past projects, review their outputs, and manage your processing history.

Installing and Running BiliNote

Quick Start

1. Clone the Repository

git clone https://github.com/JefferyHcool/BiliNote.git
cd BiliNote
mv .env.example .env

2. Initialize the Backend (FastAPI)

cd backend
pip install -r requirements.txt
python main.py

3. Initialize the Frontend (Vite + React)

cd BillNote_frontend
pnpm install
pnpm dev

Once running, navigate to http://localhost:5173 in your browser.

Required Dependencies

FFmpeg. This project relies on FFmpeg for audio extraction and format conversion.

• macOS (Homebrew): brew install ffmpeg
• Ubuntu/Debian: sudo apt install ffmpeg
• Windows: Download the binaries from ffmpeg.org/download.html. Ensure FFmpeg is added to your system’s environment variables (PATH) so it can be accessed via the command line.

CUDA (Optional). Users with an NVIDIA GPU can leverage CUDA with Fast-Whisper to significantly accelerate transcription speeds. Refer to the Fast-Whisper documentation for specific driver and toolkit requirements.

Docker Deployment

For those who prefer containerization, Docker and Docker Compose are supported. You can build and deploy the entire stack using the provided docker-compose.yml file.

Navigating the BiliNote Interface

Main Dashboard. Paste your video URL directly into the input field. From here, you can adjust audio quality settings, select your preferred AI model, and choose a specific note style. Click "Generate Notes" to begin the process.

History Tab. This section provides a log of all processed videos. You can monitor the status of active jobs, see which model was utilized for each entry, and quickly copy or export the resulting Markdown.

Model Configuration. Use this menu to add or update your AI providers. Enter your API keys and base URLs here, and use the built-in connectivity test to verify your settings before starting a new job.