The problem
In a world of AI voice cloning and trivial audio editing, "I have a recording" no longer means much. How do you prove a piece of audio is real, complete, and was captured when it claims to be — without trusting a platform to vouch for it?
What useecho does
It records audio in short chunks and turns the whole capture into a verifiable chain. Each chunk is cryptographically signed and linked to the previous one, so any edit, cut, or splice breaks the chain and is instantly detectable. Periodic witnessing anchors the chain to an external network so timing can't be faked either.
How it works
- Login with email via Privy — Nostr keys derived deterministically from your signature
- Record — audio captured in 2-second chunks
- Chain — each chunk becomes an Aqua Protocol revision, signed and hash-linked
- Witness — every 10 chunks anchored on the Nostr network
- Verify — anyone can re-check the full chain: hashes, signatures, integrity, and witnesses
Why I built it
Provenance is going to be one of the defining problems of the AI era. useecho is a bet that authenticity should be something you can prove with math and decentralized witnesses — not something you have to take a platform's word for.