We only process the data learning requires.

Audio and transcription jobs

OORA obtains the audio needed for a transcription only after the user starts that job. It is then processed by an OORA-orchestrated worker. A worker may use dedicated or on-demand GPU capacity, but it can only claim jobs already placed in the queue.

Input files are retained while a job is waiting or processing so the job can finish. After a job reaches a completed or failed terminal state, temporary input files enter cleanup and are removed within 24 hours by default.

Captions and learning data

Generated captions, translations and required processing results are stored by OORA and associated with the user’s private learning record for playback, review and shadowing features.

Transcription does not transfer copyright in the source content to OORA, and a user’s private learning results are not made public automatically.

Export and deletion

Users can request a JSON export of account data or request account deletion in OORA App settings. Account deletion starts cleanup of the associated profile, learning records and related files.

Consent records that must be retained for security, audit or legal obligations are detached from the account or de-identified and handled only under an approved retention rule.

Data required to operate the service

To maintain login, subscriptions, task orchestration, security and incident diagnosis, OORA processes information users provide and necessary technical records generated while using the service. We do not use it for public distribution unrelated to OORA’s learning service.

Contact us

For questions about privacy, data export or deletion, contact YOSHINN at support@yoshinn.com.

This page describes current product data handling. The formal policy version and effective record are the legal documents published by the OORA service.