ElevenLabs Dubbing v2 is designed to preserve more of the original speaker's performance when translating audio and video. The company says the model conditions on the original performance rather than relying only on a transcript, with support for more than 90 languages.

What ElevenLabs does

ElevenLabs introduced Dubbing v2 in May 2026 and updated the product through September 2026. The company says the system preserves tone, pacing, delivery, and emotional intent across languages.

Dubbing v2 is available through the ElevenAPI, allowing developers to integrate translation, voice cloning, dubbing, and synchronization into their own workflows.

Key capabilities to know

  • ElevenLabs introduced Dubbing v2 in May 2026 and updated the product through September 2026. The company says the system preserves tone, pacing, delivery, and emotional intent across languages.
  • Dubbing v2 is available through the ElevenAPI, allowing developers to integrate translation, voice cloning, dubbing, and synchronization into their own workflows.
  • The API supports source transcripts and target-language translations for teams that want more granular control over the generated result.
  • ElevenLabs says newer improvements include stronger handling of regional accents and better treatment of background audio, music, effects, and multiple speakers.
  • The platform's broader voice stack includes speech generation, transcription, voice cloning, and creative editing, which makes dubbing part of a larger localization workflow.

How the workflow works

AI dubbing is useful when the goal is to preserve the identity and timing of the original production rather than simply create a translated narration. The quality of the source audio and translation still matters, but the model can automate much of the production pipeline.

For publishers, marketers, educators, and software companies, this can turn one video into multiple localized versions without rebuilding every voice track manually.

  1. Provide the original media and identify the target languages.
  2. Let the system transcribe and translate, or supply controlled transcripts and translations.
  3. Generate the dubbed tracks with voice and synchronization handling.
  4. Review pronunciation, names, cultural references, timing, and any sensitive content before publication.

Practical use cases

  • Localizing product demos and software tutorials.
  • Creating multilingual educational and training libraries.
  • Publishing creator or marketing videos in additional markets.
  • Adapting documentaries and interviews for international audiences.
  • Building localized audio and video experiences into applications through an API.

What to consider before adopting it

Localization should not be treated as a fully automatic substitute for human review. Proper nouns, cultural references, legal language, accents, and brand terminology can still require editorial intervention.

Voice rights and permissions are particularly important when cloning or reproducing a speaker's voice. Organizations should document authorization and define where a generated voice may be used.

Bottom line

With an API available, Dubbing v2 is more than a creator feature: it can become part of a repeatable localization pipeline. The value is highest when teams combine automation with a defined quality-control process.