DeepL Voice

DeepL Voice for Meetings displays translated subtitles in real time during both virtual and in-person meetings, allowing each participant to read in their preferred language. The system captures audio from each speaker, processes it through DeepL's neural translation engine with an average latency of 1 to 3 seconds, and renders translated text on-screen. This mode supports multiple simultaneous languages in a single session, making it practical for multilingual all-hands meetings or international conferences without the cost of hiring multiple interpreters.

DeepL Voice for Conversations provides an alternating speech-to-translation interface optimized for two-person bilingual interactions on a mobile device. Each speaker talks in their own language, and the translation is displayed on screen and optionally read aloud for the other participant. This mode is ideal for in-person scenarios such as patient consultations, client meetings, hotel check-ins, or front-desk interactions where a full meeting setup is unnecessary.

Organizations can define how specific terms — company names, product names, industry acronyms, legal terminology — should be translated across all DeepL products including Voice. Glossaries ensure that proprietary or domain-specific vocabulary is rendered consistently, avoiding the mistranslations that commonly occur with general-purpose translation tools. Glossary entries are managed centrally and apply automatically during voice translation sessions, reducing the need for manual correction.

DeepL Voice encrypts all voice data in transit and at rest. On Pro and Enterprise plans, voice data is never stored after processing and is never used to train models. As a company headquartered in Cologne, Germany, DeepL operates under strict EU data protection regulations and offers contractual Data Processing Agreements for enterprise customers. These guarantees make DeepL Voice suitable for use in regulated industries including healthcare, legal services, and financial services where data privacy is non-negotiable.

The DeepL Voice API, generally available as of 2025, allows developers to embed real-time voice translation capabilities into their own applications, platforms, and workflows. This enables use cases such as multilingual customer support chatbots with voice input, e-learning platforms that deliver translated audio lectures, and telehealth applications that bridge language gaps between providers and patients. The API supports the same 50+ languages as the consumer product and includes glossary and security features.