UnMute works with your voice, which is personal. So this page is specific about what is collected, where it lives, and how you remove it.
Recordings you choose to keep are stored in your browser and, when saved, on our server, and you can play or delete them at any time. Speech is transcribed by Deepgram. Spoken lines from the app are synthesised by ElevenLabs, with Cartesia as fallback. Observations and generated replies are produced by Anthropic's Claude. Sign in codes are delivered by Resend. Each processor receives only what its task needs. We do not use your voice to train models, we do not sell any of your data, and there is no advertising of any kind.
Material you bring for a real event is encrypted before it is stored (AES-256-GCM). Erasing a talk is a hard erase: the content is destroyed, not flagged as deleted.
UnMute uses browser storage on your device to remember your practice. There are no tracking cookies and no third party analytics scripts.
Recordings and practice state are kept while your account is active. Delete something and it is gone from the app; server copies are removed on account deletion.
If this policy changes in a way that matters, we will say so plainly in the app before the change applies. Questions: hello@iamunmute.com.