Audio GraphRAG
Unicorn Track
A modular pipeline that ingests raw audio, applies WhisperX speaker diarization, resolves entities with Google Calendar metadata, and indexes relationships in a Neo4j temporal graph for context-grounded RAG queries. Audio GraphRAG is an end-to-end knowledge ingestion and retrieval system designed to map unstructured audio data into structured semantic networks. By combining automatic speech recognition (ASR), entity extraction, and temporal graph modeling, the project addresses the context-loss problem typical of standard vector-only RAG architectures. Developers can manage batch ingestion via batch_processor.py, run local CLI tasks using main.py, or interact with a beautiful FastAPI dashboard powered by app.py.
Category tags:Explore more applications
.png&w=96&q=75){kind=small}
Upcoming AI Hackathons
For Innovators & Creators
For Innovators & Creators
Explore more applications