ElevenLabs ships Dubbing v2 through its API, translating audio and video into 90+ languages

ElevenLabs has moved its second-generation dubbing engine out of its consumer app and into the API, giving developers programmatic control over translating audio and video while preserving how each speaker actually sounds.

What's new

According to ElevenLabs' August 10, 2026 changelog entry: "Dubbing v2 is now available through the API. It translates audio and video into more than 90 languages while preserving each speaker's voice, tone and pacing." That preservation of voice characteristics — rather than simply generating a flat text-to-speech translation track — is the core technical claim distinguishing dubbing from ordinary translated voiceover.

The API is structured around projects rather than one-shot conversion calls: "The new project-based API keeps source transcripts and translations as editable JSON." In practice that means a developer can "create a project from a file or URL, add one or more target languages, edit individual transcript segments or translations, then regenerate only the regions that changed" — so fixing a single mistranslated line doesn't require re-processing an entire audio or video file, only the affected segment.

ElevenLabs points developers to a dedicated Dubbing quickstart to create a first project, alongside reference documentation covering the project, language, and transcript endpoints that make up the new surface.

Context

Dubbing has been part of ElevenLabs' consumer-facing product for some time, primarily as a point-and-click tool for translating video content. Exposing it as a full API with project-based, editable transcripts is a shift toward treating dubbing as infrastructure that other companies can build on — localization platforms, video editing tools, and media pipelines can now integrate translated, voice-preserving dubbing directly rather than routing users through ElevenLabs' own app. It follows a pattern ElevenLabs has repeated across its product line: ship a capability in-app first, then productize it as an API once the underlying model is mature enough for third-party integration at scale.

Why it matters

For companies building localization or media-production tools, an API that preserves speaker identity across 90-plus languages — with the ability to surgically regenerate only changed segments rather than re-running a whole project — removes a real cost and workflow bottleneck compared to earlier machine-dubbing approaches that either sounded robotic or required full re-renders for small text fixes. It also puts ElevenLabs in more direct competition with dubbing and localization vendors that have historically relied on separate transcription, translation, and voice-cloning steps stitched together manually; a single project-based API collapsing those steps is a meaningful efficiency argument for switching.