Provenance first
Every piece of information keeps its source, contributor and conditions of use.
Living heritage × responsible artificial intelligence
Listen.
Connect.
Transmit.
GnawAI is a research and product initiative exploring how technology can document Gnawa recordings, connect their histories and preserve the provenance of every contribution.
Explore the prototype ↓GNAWAI / PILOT CORPUSV0.1
Explainable by design
More than recognition.
A result should show where it comes from, what it knows and what it does not.
Exact audio recognition MVP
Identify an authorised recording
Play a known recording, capture 8–15 seconds, then compare its acoustic fingerprint with the private authorised corpus.
Live microphone recognition
Listen through your microphone
The excerpt is sent securely for one-time fingerprint comparison. It is not retained as a playable public file.
Exact-recording matching only · no mlouk classification · authorised audio onlyAuthorised corpus
Your real result will appear here.
No simulated archive or hardcoded score is displayed.
The method
Trust is built.
It is never assumed.
The first phase focuses on a small, explicitly authorised corpus. No public scraping, no hidden training and no cultural claim without a source.
Human validation
A prediction remains a hypothesis until it is reviewed by a legitimate contributor.
Community governance
Artists and knowledge holders help decide what may be documented, trained on or shared.
Built with the community
AI can assist.
People decide.
GnawAI will not collect, train on or publish cultural material without appropriate permission. Its governance is designed to give artists, archives and knowledge holders meaningful control—including correction, restriction and withdrawal.
Join the pilot conversation ↗