Rimagination/bili-note

Extract Bilibili videos into learning-oriented Markdown notes with full subtitle and comment archives

What it solves

Bili Note transforms Bilibili videos and image-text posts into structured, high-quality Markdown notes for knowledge bases. It solves the problem of losing valuable information from video content by archiving subtitles, comments, and visuals, and converting them into a searchable, learnable format that avoids the pitfalls of generic summaries.

How it works

The tool extracts data from Bilibili using several layers of capability:

  • Content Extraction: It gathers subtitles (public or AI-generated via browser access), image-text body content, and comments.
  • Audio Transcription: If subtitles are missing or sparse, it uses ASR backends like Qwen3-ASR (for Chinese) or Whisper/faster-whisper (for foreign languages) via FFmpeg.
  • Visual Understanding: It can optionally extract up to 12 keyframes to create a contact sheet, allowing a vision model to interpret PPTs, code, or blackboard writing that isn't captured in text.
  • Dynamic Note Generation: It determines a "note budget" (recommended length and granularity) based on the content's information density and quality signals (like interaction rates) before writing the final Markdown note.
  • Evidence Indexing: Key claims in the notes are linked via numbered footnotes back to the original source material for verification.

Who it’s for

  • Learners and researchers using Bilibili for technical videos, courses, and long-form tutorials.
  • Users maintaining a personal knowledge base (e.g., in Obsidian) who want to archive Bilibili content for long-term retrieval and Agent-based Q&A.

Highlights

  • Quality-Aware Compression: Adjusts note detail based on the content's actual information value rather than a fixed length.
  • Multimodal Integration: Combines text, audio transcription, and visual keyframe analysis to ensure no critical information is missed.
  • Evidence-Based: Includes a materials package with original subtitles and metadata, linking notes to specific evidence indices.
  • Flexible ASR: Supports multiple transcription backends including Qwen3-ASR and Whisper.

Related

  • Project
    andrea9293/mcp-documentation-server

    A local-first document management and semantic search server for AI agents that provides a private knowledge base with a built-in web UI and hybrid search.

  • Project
    OpenBMB/BMList

    BMList is a community-maintained list of AI models with at least 1 billion parameters that have been publicly released. It offers an interactive table and visualizations to explore model trends by organization, launch time, and parameter scale.

  • Project
    liangdabiao/langgraph_multi-agent-rag-customer-support

    A multi-agent RAG customer support system built with LangGraph and LangChain that features security guardrails and a human-in-the-loop approval process for sensitive operations.

  • Project
    rupertl/eliza-ctss

    A reconstruction of Joseph Weizenbaum's original 1965 ELIZA chatbot, running on an emulated IBM 7094 computer and the Compatible Time-Sharing System (CTSS).