Voice & Biometric Data Policy
Sadorra Voice, a product of Cindara Novera, LLC ("Sadorra"), lets a person create a synthesized model of their own voice and lets their organization narrate approved content in it. A voice model is biometric information. This page says what we collect when someone enrolls, why, who processes it, how long we keep it, and how to take it back. It is the policy the consent agreement refers to, and where the two state a deadline, they state the same one.
What is collected
- Audio recordings the speaker provides during enrollment, recorded in their own browser or uploaded by them. Nobody else can upload audio for a speaker.
- A liveness recording: a short random phrase the speaker reads aloud, which is our own proof that the person consenting is the person speaking.
- The Voice Model derived from the recordings, created and stored by our technology partner.
- The consent record: the version and a cryptographic hash of the exact agreement text shown, the speaker's typed full legal name, the date and time of signing (UTC), IP address, browser or device identifiers, the uses the speaker allowed, the term end date they chose, and a PDF copy of the signed agreement.
- An audit trail of every action on the voice: invitation, consent, verification, training, each script approved and each audio file generated, and every pause, suspension, revocation and deletion. The audit trail never contains audio.
Purpose
Approved narration only. A Voice Model speaks the exact text of a script that has passed Sadorra's approval process, and by default the speaker personally approves every script. Nothing lets a voice say free-form text, and there is no real-time or conversational use. All audio generated with a Voice Model is labeled as AI-generated, both as a spoken notice at the start and as a visible label wherever Sadorra plays it.
Service providers
Voice Models are created and hosted by ElevenLabs, Inc. under its Professional Voice Cloning service. Recordings go to ElevenLabs to create the model; ElevenLabs stores the model and generates the audio. The ElevenLabs account Sadorra uses has the setting that would allow ElevenLabs to improve its models with our data turned off, so recordings, models, scripts and audio sent through Sadorra are not used to train ElevenLabs' models. Neither Sadorra nor the speaker's organization uses recordings or Voice Models to train, retrain or improve AI models beyond creating and running that speaker's own Voice Model.
Retention schedule and destruction
- Raw audio: Source recordings are permanently deleted from Sadorra's systems within 24 hours after the Voice Model is successfully trained.
- Voice Model retention: The Voice Model is kept only until the earliest of: (1) the term end date the speaker chose when consenting; (2) the effective date of the speaker's revocation; or (3) the termination or expiration of the organization's agreement with Sadorra.
- Destruction deadlines:
- On revocation, Sadorra will disable the Voice Model and send a deletion request to ElevenLabs, Inc. within 1 hour.
- On term end or contract termination, Sadorra will do the same within 30 days.
- Copies held in Sadorra's or ElevenLabs' backup systems are purged under each party's backup retention schedule, no later than {backup purge window — confirm with ElevenLabs} after the deletion request.
- Confirmation: Sadorra will email the speaker written confirmation once deletion is complete.
- Consent records and the audit trail are kept for the consent term plus three years, so that what was agreed and what was done with the voice can be shown afterwards. They contain no audio.
- Suspension is not deletion: a suspended voice cannot generate audio, and nothing is removed until the speaker revokes or the term ends.
How to revoke, and how to request deletion
A speaker may revoke consent at any time from their Sadorra account settings, where the control is always visible, or by emailing privacy@sadorra.app. Revocation stops all new audio at once and starts the deletion described above. Audio already published before revocation may remain in use unless otherwise agreed in writing with the speaker's organization. A speaker may also ask us to delete their consent record once the retention period above has ended, and may ask at any time for a copy of what we hold about their voice.
No sale, no profit from voice data
Sadorra does not sell, lease, trade, license for third-party profit, or otherwise profit from recordings, liveness clips or Voice Models. Voice data is used only to provide the service described on this page to the speaker's organization.
Who can reach voice data
The liveness recording and the signed consent PDF are stored encrypted and can be read only by the speaker and by Sadorra's platform administrators investigating misuse. Provider identifiers for a voice are never shown to anyone in the organization. Enrollment, approval and generation are available only through the Sadorra application, never by API key.
Contact for privacy requests
Email privacy@sadorra.app. Cindara Novera, LLC is a Georgia limited liability company in the United States. For how the sadorra.app website itself handles information, see the privacy page.