Voice privacy notice
This notice covers voice.globalieta.com and its connected Voice API. Updated 8 October 2026. It describes current product behavior; it is not a certification of compliance.
Operator and contact
AURA Voice is operated by GLOBALIETA EDUCATETECHNOLOGY PRIVATE LIMITED, also identified as IETA. CIN: U62011GJ2026PTC183275. The company's incorporation certificate, dated 5 September 2026, records its mailing address as A2, 103 Marion Residency, Dumas, Surat, Gujarat, Dumas, Choryasi, Surat- 394550, Gujarat. This is the address in that dated record; changes since incorporation have not been independently verified. The published support contact is support@globalieta.com. For a privacy request, use the subject “AURA Voice privacy request”. Do not email passwords, API secrets or voice recordings. A designated grievance officer and confirmation that the privacy mailbox is monitored are still pending.
What Voice processes and why
Voice uses your signed-in account identifier to keep work account-scoped. Submitted text, speech settings, language choices and authorized recordings are processed to generate, convert, transcribe or dub audio. Saved voices, permission confirmations, listening approvals, job status, project scenes, pronunciation entries and voice memories support your private workspace. Usage and technical records support quota enforcement, troubleshooting and security.
Recordings and voice permission
A voice recording can identify a speaker. Upload only material you own or have explicit permission to process. Clone permission does not authorize public publication or every use of a person's voice. Permission checkboxes are not restored automatically. Unsubmitted recordings stay in browser memory; refresh or leaving the workspace requires selecting them again. Submitted recording sources may be stored with the job or private voice.
Infrastructure and access
The current primary Voice host runs in Google Cloud's us-central1 region in the United States. Voice inference runs on operator-managed services. “Sovereign” does not mean that all data is stored in India. Generated audio and workspace APIs require authentication and account ownership checks. Public voice publication is disabled. Private audio is not used for training by the current Voice product pipeline.
Realtime speech and expressive conversation
Realtime speech currently uses English preset voices. Expressive conversation is an English Preview and requires your explicit permission before microphone processing or a typed conversation. Microphone audio is processed by operator-managed speech services and is not saved to disk by the conversation pipeline. A transcript and recent turns are held temporarily in server memory to prepare replies. Conversation memory expires after ten minutes of inactivity; End conversation requests clearance, and the interface confirms when server clearance succeeds. Reply preparation can take tens of seconds.
Acoustic tone and emotion cues are uncertain estimates, not a diagnosis or a reliable statement about how a person feels. Guided playback uses the chosen speech tone; Fast streaming does not support tone controls. Generated reply text and audio may create private Voice generation records and appear in your history. Ending a conversation does not erase those saved generation records or external records; their removal follows the applicable Voice request and retention controls described below. Preview availability does not establish native listening quality or an approved retention schedule for every storage location.
Browser storage
Authenticated speech and supported text drafts use storage scoped to this browser tab and account. Session authentication and public application assets are required to operate the service. The service worker caches public interface files and an offline help page; private API responses, uploaded recordings and generated audio are excluded. Optional IETA visitor analytics and the external sales-chat launcher are not loaded by this Voice release. Cookies set previously by other IETA products may still exist; this change does not erase records held by those products. Shared devices still require signing out and closing your private tabs.
Retention and deletion limits
Settings displays an aggregate inventory and a separate Voice data-request workflow. The older cleanup planner remains audit-only; a completed-retention value of zero preserves completed generations indefinitely. Permanent erasure requires an operator-approved schedule for product records, security records, backups and request receipts. This schedule, the designated grievance contact and a response period still require confirmation. Until then, submitting an erasure request does not delete data.
Your requests and choices
Open Privacy in Settings to create an access request and download your owner-scoped Voice records, stored recordings and generated audio as a private ZIP. Credentials and internal file-path fields are excluded. You can also submit a correction, grievance or erasure request and track its reference and operator response. Access downloads cover known Voice storage; unknown categories are disclosed. Infrastructure logs, backups, shared sign-in and other products require separate handling.
How permanent erasure works
Once the operator confirms the policy, review the displayed scope, sign in again if required and type the exact confirmation phrase. The plan expires after 15 minutes and must still match your records and files. Active or unverified jobs, unclear ownership, unknown storage and an unelapsed minimum retention period block deletion. The workflow deletes and verifies live Voice records and files, then keeps the request open until the operator records dispositions for backups, infrastructure logs, shared identity, shared project derivatives, retained records and streaming memory. Retention requires a stated basis and expiry; it is displayed separately. Live deletion alone is not complete account erasure. Operator receipts document external handling and do not independently certify a provider's deletion. Backup restores require reconciliation before access reopens.
Other products and changes
The parent IETA website has its own notices. They do not replace this product-specific description. External support and account services have their own data handling. Material changes to Voice processing require the operator to update this notice and the relevant product controls.