When you use DeepL Voice, your spoken audio is processed live and neither the audio recording nor the text transcript is saved after the session ends.
This analysis describes what DeepL's agreement states, permits, or reserves. It does not constitute a legal determination about enforceability. Regulatory applicability and practical outcomes may vary by jurisdiction, enforcement context, and individual circumstances. Read our methodology
Voice translation involves the real-time capture and processing of spoken audio, which may include sensitive or personally identifiable speech; the policy's no-storage claim is significant for users concerned about audio surveillance or transcript retention.
Interpretive note: The policy does not specify the technical infrastructure or subprocessors used in real-time audio processing, creating uncertainty about whether audio data is fully contained within DeepL's own systems during transit.
The updated terms establish explicit procedures for three new account-based storage and processing features. Translation Memories allow users to store source and target language segments for reuse; Translation Memory Generation processes uploaded document pairs to generate Translation Memories, then deletes the source documents immediately; and Adaptive Translation uses stored Translation Memories and reference documents to customize translations, with numeric representations (embeddings) calculated and stored in the user's account. The policy states that all content and embeddings are used only within the user's own account, never shared with other customers, and never used to train or improve DeepL's models. The policy specifies GDPR Article 6(1)(b) as the legal basis for processing (contract performance). You can deactivate Adaptive Translation in the relevant Style Profile at any time.
View change record →Users of DeepL Voice can expect, according to the policy, that their audio and transcripts are not retained after translation, which is a privacy-favorable design for spoken communication. However, users in jurisdictions with strict wiretapping or biometric data laws should be aware that real-time audio processing may engage additional legal requirements.
How other platforms handle this
to withdraw your consent to our processing of your data (where such processing is based on consent)
Request Restriction or Object to Processing of your information.
If you choose to reveal any personal information about yourself to other users, you do so at your own risk. We strongly encourage you to use caution in disclosing any personal information online.
"DeepL Voice processes audio data in real time. We do not store audio recordings of your conversations. The transcribed text is also not stored after the translation has been completed.Excerpt from DeepL's Privacy Policy
(1) REGULATORY LANDSCAPE: Real-time audio processing engages GDPR's special category data provisions if speech processing involves biometric identification, as well as applicable wiretapping and interception laws.
Enforcement risk, jurisdiction flags, contract triggers, and due diligence action items.
Ad personalization controls removed. Contact scanning added. Advertiser data partnerships quietly dropped. A timeline of every change.
Get the research letter
Companies change their terms quietly. We read every version and catch what actually changed. One email a week on the changes that matter and what they mean.
Voice translation involves the real-time capture and processing of spoken audio, which may include sensitive or personally identifiable speech; the policy's no-storage claim is significant for users concerned about audio surveillance or transcript retention.
Users of DeepL Voice can expect, according to the policy, that their audio and transcripts are not retained after translation, which is a privacy-favorable design for spoken communication. However, users in jurisdictions with strict wiretapping or biometric data laws should be aware that real-time audio processing may engage additional legal requirements.
ConductAtlas has identified this type of provision across 290 platforms. See the full comparison.
No. ConductAtlas is an independent monitoring service. We are not affiliated with, endorsed by, or sponsored by DeepL.