ElevenLabs released Dubbing v2 in May 2026 — and the shift is subtle but important. Instead of translating a transcript and generating neutral TTS, the model conditions directly on the original performance: tone, emotion, pacing, and delivery. For brands producing campaign films, that means localization can preserve the energy of a hero spot rather than flattening it into generic dubbed audio.
What actually changed
- Performance-first dubbing — emotion and acting carry into each target language
- Sync-aware translation — natural reformulations aligned to video timing
- One-click workflow in ElevenCreative for creators; ElevenProductions for pro pipelines
- 90+ output languages with starts/stops matched to the source clip
Our test results
We ran a 45-second source clip through the free tier. The exported video came back compressed and pixelated — so for production use, we recommend extracting the dubbed audio and transcript, then re-attaching them to your original master video in DaVinci or Premiere. The dubbing quality itself was excellent: voice cloning captured delivery faithfully, including speech mannerisms. This is audio dub only — no lip sync.
Verdict
AI dubbing was never flat by accident — it was reading a transcript. Dubbing v2 listens to the performance. That changes everything for campaign localization.
For fashion, beauty, and luxury clients running global campaigns, Dubbing v2 is a practical lever: one master film can ship in EN, ES, DE, and beyond without rebuilding the entire production. The emotional fidelity is what separates this from earlier AI dub tools — educational and narrative content finally travels with the same energy as the original.
Available now at elevenlabs.io/dubbing. API access is coming soon; ElevenProductions offers a human-in-the-loop path for broadcast-grade deliverables.