DeepL launches voice translation that preserves speaker identity

DeepL has announced significant advancements to its DeepL Voice technology, enabling real-time voice translation that preserves the speaker's unique identity, including tone, emotion, and rhythm. While voice-to-voice translation is available in over 30 languages, the technology can specifically maintain the original characteristics of the speaker's voice in 14 languages.

To facilitate multilingual communication, the company is launching a dedicated desktop application designed to work alongside online meeting platforms such as Zoom, Microsoft Teams, and Google Meet. This allows users to participate in multilingual conversations without relying on browser tabs or native platform integrations.

A study by the firm Slator previously ranked DeepL Voice as a top performer, noting quality scores of 96.4/100 for Zoom and 96.3/100 for Teams. The report indicated that the technology reduces critical error rates by an average of 76% compared to competing solutions, with 96% of professional linguists preferring the tool during blind tests.