Data handling
What the KugelAudio core API stores, for how long, where it is processed, how to delete your voices, and how to get a data processing agreement.
On this page
This page covers the text-to-speech API.
What is stored
| Data | What is kept | How long |
|---|---|---|
| Request logs | One row per authenticated text-to-speech request: time, outcome and error code, endpoint, model, voice ID, API key, audio duration, character count, cost, trace ID | Up to 14 days; this is what the Logs page shows. See Find a request in your logs |
| Voices you create | The reference audio, the generated sample, the avatar and cached encodings | Until you delete the voice or a reference |
Requests rejected for a missing or invalid API key are not in the request logs.
Delete your data
Delete a voice to remove its reference audio, sample, avatar and cached encodings for good. You can also delete a single reference. A published voice cannot be deleted.
Where requests are processed
The default endpoint can answer from outside the EU. To keep a request in Europe, use the EU endpoint. See Regions. Organizations on the EU-Only Hosting data processing agreement must send every request to it.
Data processing agreement
The Trust Center has KugelAudio's documents, controls and subprocessors. For a data processing agreement, write to hello@kugelaudio.com.