No audio recording.

Beta: transcripts are kept while we tune translation quality

Verata is in an early quality phase. During it, the text transcript of every session is stored, so that translation errors can be found and corrected against real speech. This is text only — the no-audio rule below is unchanged.

If you did not switch transcript capture on for a session, that transcript does not appear in your account. It is used solely to correct the service, it is deleted automatically after 30 days like any other, and the whole set collected during this phase is deleted when the phase ends.

This section will be removed, and capture returned to opt-in only, once the phase is over.

No audio recording

Verata never records or stores the conversation audio. Speech is streamed to our recognition provider only to produce the live translation, moment to moment, and is then discarded. There is no audio file to leak, subpoena, or lose.

Subtitles are ephemeral

The on-screen text lives only in memory for the length of the session — a short rolling context of recent lines keeps the translation coherent — and is gone the moment the session ends.

Transcripts in your account

Turning on transcript capture for a session is what puts that transcript in your account, where you can read it. It is stored on the platform and automatically deleted after 30 days. It is per session, and always visible to you.

During the beta phase above, transcripts are also kept for quality work when this is off — those are never shown in your account.

We keep only what runs and bills the service

Your account and per-minute usage totals — enough to operate and invoice. Nothing about the content of a conversation, unless transcript capture is on or the beta phase above applies.

Processing partners

Recognition, translation and voice run through named providers (Soniox, Anthropic Claude, Microsoft Azure, OpenAI) that process the stream transiently to return a result and, under contract, do not use the entrusted data to train their models. See the Privacy Policy for details.

Why it matters

The design follows data minimization and purpose limitation under PIPA, so a clinic can bridge a language gap without creating a new store of patient conversations.