How vaishnavimd/ai-meet-summerizer-using-rag- works
AI Meeting Summarizer turns uploaded recordings into searchable transcripts, meeting Q&A, and structured minutes. The principal workflow runs from browser upload through Flask job orchestration, video/audio conversion, Whisper transcription, transcript chunking and embedding, then persists meeting data for results. Separate request paths retrieve relevant chunks for Q&A or generate MOM documents; the React results view presents the minutes and supports follow-up questions. The README documents these capabilities; several backend and UI excerpts are partial, so uncertain internal wiring is omitted.
Main entry points: app.py Flask API · pipeline.py Pipeline orchestrator · qa_engine.py Retrieval and MOM · App.jsx App state flow
Components 21
Meeting Interface
- React entry
frontend/meeting-summarizer-integrated/src/main.jsx - App state flow
frontend/meeting-summarizer-integrated/src/App.jsx - Video upload
frontend/meeting-summarizer-integrated/src/components/FileUpload.jsx - Job progress
frontend/meeting-summarizer-integrated/src/components/ProcessingLoader.jsx - Results dashboard
frontend/meeting-summarizer-integrated/src/pages/Dashboard.jsx - Minutes display
frontend/meeting-summarizer-integrated/src/components/MomDisplay.jsx - Meeting Q&A chat
frontend/meeting-summarizer-integrated/src/components/ChatBox.jsx
API and Jobs
- Flask API
app.py - Job registry
app.py - Pipeline orchestrator
pipeline.py
Recording Ingestion
- Video conversion
video_to_audio.py - Speech transcription
transcriber.py - Chunk and embed
chunk_embedder.py
Meeting Analysis
- Retrieval and MOM
qa_engine.py
Other components
- Meeting user
- FFmpeg
- Whisper
- Ollama service
- Embedded chunks
- Transcript data
- MOM document
Connections 29
- Meeting user → React entry: uses
- React entry → App state flow: renders
- App state flow → Video upload: shows upload
- Video upload → Flask API: uploads video
- Flask API → Job registry: tracks job
- Flask API → Pipeline orchestrator: starts ingestion
- Pipeline orchestrator → Video conversion: converts video
- Video conversion → FFmpeg: uses
- Pipeline orchestrator → Speech transcription: transcribes audio
- Speech transcription → Whisper: transcribes with
- Pipeline orchestrator → Chunk and embed: builds and embeds
- Chunk and embed → Ollama service: requests embeddings
- Chunk and embed → Embedded chunks: persists chunks
- Speech transcription → Transcript data: writes transcript
- Flask API → Retrieval and MOM: generates MOM
- Pipeline orchestrator → Retrieval and MOM: answers or generates
- Retrieval and MOM → Embedded chunks: retrieves chunks
- Retrieval and MOM → Ollama service: requests generation
- Retrieval and MOM → MOM document: writes MOM
- Flask API → Job progress: serves job status
- Job progress → Flask API: polls status
- Flask API → App state flow: returns results
- App state flow → Results dashboard: shows results
- Results dashboard → Minutes display: displays minutes
- Results dashboard → Meeting Q&A chat: shows Q&A
- Meeting Q&A chat → Flask API: asks question
- Flask API → Job registry: resolves meeting
- Flask API → Pipeline orchestrator: requests answer
- Flask API → MOM document: serves download