Skip to content

Latest commit

 

History

History
33 lines (21 loc) · 1.94 KB

File metadata and controls

33 lines (21 loc) · 1.94 KB

🎓 Smart Classroom

The Smart Classroom project is a modular, extensible framework designed to process and summarize educational content using advanced AI models. It supports transcription, summarization, mindmap generation and future capabilities like video understanding and real-time analysis.

The main features are as follows:

  • Audio transcription with ASR models (e.g., Whisper, Paraformer)\
  • Speaker diarization is supported using Pyannote Audio models.\
  • Summarization using powerful LLMs (e.g., Qwen, LLaMA)\
  • MindMap Generation using Mermaid.js for visual diagram rendering of the summary\
  • Plug-and-play architecture for integrating new ASR and LLM models\
  • API-first design ready for frontend integration\
  • Video analysis

Get Started

To see the system requirements and other installations, see the following guides:

How It Works

The basic architecture follows a modular pipeline designed for efficient audio summarisation. It begins with audio preprocessing, where FFMPEG chunks input audio into smaller segments for optimal handling. These segments are processed by an ASR transcriber (e.g., Whisper or Paraformer) to convert speech into text. Finally, an LLM summariser (such as Qwen or Llama), optimised through frameworks like OpenVINO IR, Llama.cpp, or IPEX, generates concise summaries, which are delivered via the output handler for downstream use.

High-Level System Diagram

For more information see How it works

Learn More