Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.cartesia.ai/llms.txt

Use this file to discover all available pages before exploring further.

Overview

ZDR is part of Cartesia’s broader enterprise security posture. Cartesia is GDPR, SOC 2 Type II, PCI-DSS service provider, and HIPAA compliant. For full details, visit our Trust Center. You may opt into Zero Data Retention if you require that we do not retain your content. Content means the customer-provided or model-generated text/audio payloads.

What is covered by ZDR?

DataDescriptionRetained?
Text input (TTS)The transcript or text you send to generate speech Not retained
Audio output (TTS)The generated speech audio returned by the API Not retained
Audio input (STT)The audio you send for transcription Not retained
Transcript output (STT)The transcribed text returned by the API Not retained
Operational metadataRequest identifiers, usage totals, account information, and service health data Retained
ZDR applies to TTS and STT inference only. ZDR is not available for voice cloning, PVC, voice creation, or other workflows where customer-provided source material must be retained or processed differently. We may retain non-content operational metadata needed to run the service, including request identifiers, usage totals, account information, and service health data.

FAQs

You must be on an enterprise plan to enable ZDR. Go to play.cartesia.ai/subscription to see what plan you are on and make changes.
Yes. All models available through Cartesia’s TTS and STT APIs are covered.
No. ZDR changes retention behavior, not model behavior.
Because data is not retained, Cartesia cannot access or recover request content after the request completes. Product features that depend on retained content, such as Playground history, may not be available for ZDR-enabled requests. Troubleshooting may also be more limited, so support may ask you to reproduce the issue, share request metadata, or provide sanitized examples.
They require retaining uploaded audio samples or derived voice artifacts to function.
No. Once your organization admin enables it on the Cartesia Playground on your Enterprise account, ZDR applies to all eligible API requests automatically.