Voice-Passphrase Retrieval
Speak a secret phrase — or type it — to retrieve and decrypt your message. The system converts speech to text entirely on your device using the Web Speech API.
Create encrypted messages retrieved by speaking a secret passphrase. Voice processing occurs locally on your device — only a hash of the resulting text is sent for verification. PBKDF2 key derivation with 600,000+ iterations. Then it burns.

Watch the EchoDrop demo: how a spoken passphrase protects an encrypted message from creation through retrieval — with speech recognition that never leaves your device.

EchoDrop turns a spoken passphrase into a decryption key — with voice processing designed to stay on your device. Proven cryptographic standards meet on-device speech recognition.
Speak a secret phrase — or type it — to retrieve and decrypt your message. The system converts speech to text entirely on your device using the Web Speech API.
Voice processing occurs locally on your device using the Web Speech API. No audio recordings, voice data, or speech patterns are transmitted to City of Hats servers — only a text hash is sent for verification.
Your passphrase is stretched through PBKDF2 with a high iteration count (600,000+) to derive the AES-256-GCM encryption key. Only a SHA-256 hash is stored server-side for matching.
Set EchoDrops to burn after first read, limit retrieval count, add time-lock delays, or auto-destroy after a countdown. Once burned, the encrypted payload is permanently deleted.
From compose to speak — the passphrase is never stored. Only the hash touches the server.
Write your secret message. Set lifecycle rules — burn-after-read, time-lock, retrieval limit, auto-destroy.
Choose a system-generated passphrase or create your own. PBKDF2 derives the encryption key. Only the SHA-256 hash is stored.
Tell your recipient the passphrase — in person, by phone, or through a separate channel. No digital link needed.
The recipient speaks the passphrase into City of Hats. On-device STT converts it to text, the hash matches, and the message decrypts. Then it burns.
When the retrieval method itself needs to be untraceable — no links, no files, no digital trail.
Share a spoken passphrase in person. The source speaks it later to retrieve your encrypted instructions. No digital link to intercept or subpoena.
Share sensitive directives retrieved by voice. No links in browser history, no files on disk, no messages in chat logs. The passphrase exists only in memory.
Share sensitive compliance decisions or trade instructions via voice-triggered retrieval. Full audit trail of lifecycle events without exposing content.
Deliver privileged legal communications that can only be retrieved by speaking the passphrase. No forwarding, no screenshots of links — just a burned message.
Coordinate breach response through voice-triggered instructions. Share the passphrase on a secure call. The response team speaks it to get the playbook. Then it's gone.
Share HIPAA-sensitive patient data between practitioners using a spoken passphrase. No digital trail — the message self-destructs after retrieval.
Every architectural decision is designed to eliminate digital traces of both the message and the retrieval method.
Compose a message. Set a passphrase. Share it verbally. The recipient speaks it to decrypt. Then it's gone forever.