Speak your own language, hear theirs: ElevenLabs opens realtime voice translation to enterprise callers
Synopsis
ElevenLabs released Eleven Realtime Translation, a live voice translation model that lets each person speak their own language and hear the other in theirs, in a natural, expressive voice, while the conversation happens. Previously, a customer calling in one language often waited for an agent who understood them, or an interpreter had to join the call. The model runs on the low-latency speech model Eleven v4 Turbo, covers more than 90 languages with automatic language detection, keeps one voice consistent across languages, and is in alpha for ElevenAPI enterprise customers.
Interpretation
Cross-language calls move from finding an agent who speaks the caller's language to each side speaking their own language and hearing the other's in real time, with translation happening during the conversation rather than afterward. Previously, a customer calling in one language often waited on hold for an agent who could understand them, and often had to deal with questions in a language they did not speak as fluently. Deutsche Telekom is already using Realtime Translation to build new calling features in its Magenta AI Call Assistant, and its VP Consumer AI, Lena Drubel, described it as a complete new era of how voice can connect people.
The translation is performed by a speech model that keeps one voice through the session; users pick any of more than 17,500 voices in the library or build an Instant or Professional Voice Clone, and that voice stays consistent for the entire session. It uses Eleven v4 Turbo by default, a low-latency speech model optimized for realtime applications, and a single voice performs more consistently across multiple languages so a brand or agent sounds consistent in every market. The model is available in more than 90 languages with automatic language detection, and teams set how quickly translation starts, from higher-quality translation for complex calls such as billing or claims to the lowest latency for faster back and forth.
Perspective
The report covers the product information ElevenLabs published: Eleven Realtime Translation is available in alpha for ElevenAPI enterprise customers, uses Eleven v4 Turbo by default, supports more than 90 languages with automatic language detection, and draws voices from a library of more than 17,500 or from Instant or Professional Voice Clones. The stated settings are telecom operators embedding it in the calling layer, contact centers placing it between agent and caller, and live events where each attendee hears the speaker in their own language.
The public material gives no translation accuracy, end-to-end latency figures or concurrency scale, and does not describe differences in language-pair coverage during alpha. Deutsche Telekom's statement describes calling features being built, not a completed deployment result.