No, no, and no. Audio streams directly from your browser to the voice provider and is never stored on Profiled’s servers; there are no voiceprints, no voice identification, and no emotion inference. Transcripts — the text of what was said — are kept for 12 months, and you can delete them any time.
Detail
Where the audio goes: voice conversations stream browser-direct to the voice provider. The audio does not transit Profiled’s servers, and it is not recorded there. On the failover path, audio is relayed in flight and likewise not stored — the sub-processors page describes both routes.
What is never done with it: Profiled builds no voiceprint, does not use voice to identify you, and runs no emotion inference on how you sound. These commitments are stated in the AI Disclosure Statement on the site, which is the authoritative version of this answer.
What is kept: transcripts. The text transcript of a voice session is retained for 12 months and then purged automatically. You can delete transcripts yourself at any time from Settings → Data, individually or all at once — deletion does not wait for the 12-month clock.
No model training. The providers that process voice do so under agreements that prohibit using your data to train their models, and the optional data-sharing settings on Profiled’s provider accounts are disabled.
If a voice session will not start at all, that is usually a browser microphone permission issue rather than anything on this page — see the troubleshooting article below.
Related
- How long are voice transcripts kept, and can I delete them?
- Voice says "Could not access your microphone."
- How do voice minutes work, and what happens when I run out?
Last verified: 2026-08-18