Your recording.
A clear account of its handling.

Account and access

You sign in with ChatGPT. OpenEar receives a site-specific account identifier and sign-in identity, including your email, to associate recordings and service requests with you. We do not collect a separate password.

Originals and restored audio are stored in private cloud object storage. Metadata, transcripts, quotes, and observations are stored in the site database. Sites hosts the service with Cloudflare storage. Your recordings are not published in a public gallery.

Your account can access its own recordings. Submitting a service request authorizes the designated OpenEar operator to access and download that recording, perform restoration on a local workstation using tools such as iZotope RX, and return audio and transcripts. The operator can see your account email for service correspondence. Hosting providers may have operational access. This is not end-to-end encrypted storage.

Free sample and paid service

The server validates WAV structure and calculates a SHA-256 checksum. A person reviews submitted recordings, selects an excerpt, and returns cleaned audio and a transcript when useful. A quote states the amount, currency, and scope of further work. Accepting a quote requests an invoice; the site does not charge a card or automatically send an invoice. The operator arranges payment separately and confirms it before releasing paid work.

Restoration and transcription can be mistaken. Uncertain or unintelligible passages may remain. The service does not certify evidence, identify a crime or a person’s intentions, diagnose anyone, or establish an authenticated chain of custody. A checksum identifies bytes.

Optional learning permission

Service permission and research permission are separate. An unchecked optional box lets you allow your recording, restoration settings, operator notes, and before-and-after examples to be retained for internal evaluation and development of restoration automation. Declining does not prevent service. This permission does not authorize public release or third-party model training.

Use “Withdraw optional learning permission” on a request to revoke it and remove its saved website learning records. This does not cancel restoration or delete service deliverables. Local RX exports and copies require operator removal: contact support to request their deletion. Developing an automation or training integration requires further work; logging a restoration step does not automatically train a model.

Research workspace and cloud processing

The separate listening lab preserves original audio and user observations. An operator-only research preview can send a separately confirmed recording to Modal workers for enhancement and transcription comparisons. Public service uploads do not automatically invoke that research preview. Customer audio is not automatically sent to transcription model providers by the manual service workflow.

Retention and deletion

Recordings remain in active website storage until deleted. In the listening lab, use “Delete recording and observations” to remove the source and its website observations, preview derivatives, service request, learning records, and service audio. Download anything you wish to retain first. For local operator copies, completed project exports, account or billing records, or other privacy requests, contact support. Operational and financial records may need separate handling; we will explain what can be removed.

Deletion does not promise immediate removal from provider backups or operational logs. An audio-free research admission ledger can remain to enforce processing limits. A cloud job already running may continue until its timeout; cancellation prevents further publication of its results.

Upload limits and authority

WAV uploads are limited to 100 MB each, 20 recordings and 250 MB per account, with an upload rate limit. Contact support about larger recordings or other formats before sending them. You must have authority to upload and process the recording. This statement does not replace the recording-consent requirements applicable to your situation.

Contact

For service and privacy requests, email info@hddfwrestoration.com. OpenEar is not monitored for emergency response.