Why start with a voice?
A small experiment in making online identity a little less disposable.
14 September 2026 · Introduction
A display name is easy to change. So is an email address. For a small community, that makes a familiar moderation problem surprisingly persistent: someone leaves, changes the label, and starts over.
OpenVoiceGate explores a different entry point. Read a fresh phrase. The verifier looks for a returning voice profile, or creates one when it finds a clear non-match. Choose a name or keep it hidden, then enter the site. New visitors and returning visitors take the same path.
What connects voice and community?
Continuity. A recognized profile can keep its site identity when its display name changes. That gives a site something more persistent to moderate than a freshly typed username. For a visitor, it means no email or password is needed to get started.
That connection is useful only if matching works well enough and the effort is worth it. Speaking for 15–30 seconds is not automatically easier than using a passkey. For some people and places it will be the wrong tradeoff.
The obvious question: can a bot get in?
Yes. Synthetic voices have passed this experiment. A person can also sign in and then use automated tools to write messages. A fresh voice phrase, repeat-profile matching, spoof detection, and message moderation answer different questions.
Our goal is fewer throwaway accounts and better community continuity. We have not established a human-only space, and a lively spectrum on the recording screen does not change that.
Your voice is not just another username
The current verifier receives the recording and stores an encrypted voice template. Participating sites get their own identifier and optional names you approve, but the verifier can link visits. This is an explicit privacy tradeoff, not a claim that personal information never leaves your device.
The public release contains the integration helper, site, and example application. It gives developers a concrete flow to try and inspect at the site boundary. The hosted verifier is separate.
Why release it now?
Because a working proof of concept is a better conversation than a large promise. One human desktop smoke test worked; a phone attempt exposed a phrase-check failure. That is enough to invite focused feedback, not enough to call the identity problem solved.
We’re starting with independent chats and forums. Try the entry flow. Read the integration. Tell us where the friction, privacy cost, or recognition behavior makes it unsuitable—and where continuity would be useful.