How long does Audexum keep my audio, and is it used to train AI?
Audexum keeps generated audio for 30 days and history records for 90 days, deletes audio you upload for transcription as soon as it is transcribed, and does not use your inputs or outputs to train, fine-tune or evaluate models.
How long is each kind of data kept?
- Generated audio files: 30 days, then deleted.
- History (your text and settings): 90 days.
- Dictation transcripts, batch jobs, audiobook and dubbing jobs: 90 days.
- Audio sent for transcription: the working copy is deleted as soon as transcription finishes. The transcript text is stored.
- Cloned voices: until you delete them. The reference recording is deleted 12 months after enrollment.
- Public share clips: no expiry, for as long as they exist. You can delete your own.
Is my data used to train AI?
No. Audexum does not use your inputs or outputs to train, fine-tune or evaluate models, and does not sell personal data.
Where is it processed?
Audexum's servers are its own hardware in Bulgaria. A small number of third parties are involved, for payments, email, sign-in, the edge network, and hosted AI models through OpenRouter for the V4 voice engine and when Audexum's own GPUs are unavailable. All are named in the Privacy Policy and on the sub-processors page.
To delete everything, see How do I delete my account or download my data?
Связанные материалы