AnkiLM

An automated "Import-to-Learn" pipeline that transforms raw lecture materials into structured, exam-ready study artifacts.

🚀 Overview

This project is a sophisticated study assistant designed to automate the grunt work of university studies. Instead of manually transcribing lectures or summarizing endless slide decks, this system ingests raw data (audio recordings & PDFs) and uses an AI-powered pipeline to generate concise summaries, deep-dive explanations, and spaced-repetition flashcards.

Information flows from raw files into a local orchestration layer, through an AI processing engine (Whisper & Dify), and finally renders into this interactive Frontend Dashboard for review and study.

✨ Key Features

1. The "Import to Learn" Pipeline

• Audio Ingestion: Automatically detects and processes lecture recordings (MP3).
• High-Fidelity Transcription: Uses OpenAI Whisper to generate accurate text from speech.
• Smart PDF Integration: "stamps" slide decks with page numbers, allowing the AI to cite specific slides in its explanations.

2. AI-Powered Analysis (Dify Workflow)

The system feeds prepared data into a Dify workflow to produce structured artifacts:

• 📄 Summaries: Executive summaries of the lecture topics.
• ✨ Refined Transcripts: Cleaned-up text, removing filler words and stuttering.
• ⚡ TL;DRs: One-page cheat sheets with key exam topics.
• 📖 Concepts & Definitions: Extracted terminology tables.
• 🧮 Example Problems: Breakdown of practical examples mentioned in class.
• 🧠 Anki Cards: Formatted flashcards ready for import into spaced repetition tools.

3. Interactive Frontend Dashboard

This repository hosts the modern web interface used to study the generated content.

• Tech Stack: Built with React 19, Vite, Tailwind CSS v4, and Framer Motion.
• MDX Integration: Renders rich text content directly from the generated markdown files.
• Responsive UI: A clean, distraction-free reading environment.

🛠️ System Architecture

graph TD
    A[Raw Input] -->|MP3 & PDF| B(Orchestrator)
    B -->|Audio Processing| C[Whisper AI]
    B -->|Slide Processing| D[PDF Stamper]
    C & D --> E[AI Brain / Dify]
    E -->|Structured MDX| F[Study Artifacts]
    F --> G[Frontend Dashboard]

Loading

1. Orchestrator: A local script watches subject directories for new files.
2. Processing: Audio is transcribed; slides are indexed.
3. Synthesis: The Dify agent analyzes the content against the slides.
4. Presentation: This React application renders the final 01-summary.mdx, 06-anki.mdx, etc.

📸 Screenshots

💻 Tech Stack

Frontend:

• 
• 
• 
• Framer Motion (for smooth transitions)
• MDX (for content rendering)

Pipeline / Backend:

• Deno (Orchestrator)
• OpenAI Whisper (ASR)
• Dify (LLM Workflow Orchestration)

📝 Usage

• Process New Content: (Backend step) Drop files into the monitored data folder and run the orchestrator.
• Study: Open this dashboard. Navigate to the specific lecture to view the auto-generated summaries, definitions, and flashcards.

🔜 Future Improvements

• Real-time processing status updates in the dashboard.
• Direct PDF viewer integration alongside the transcript.
• Export to Notion integration.