ElevenLabs has launched Dubbing v2, an AI model designed to improve voice translation quality. Unlike the previous version that relied on text transcripts, the new model processes original audio directly to preserve the speaker's tone, pacing, and emotional delivery.
The system features sync-aware translation, which automatically aligns speech timing with the source audio. This process removes the need for manual lip-sync adjustments and preserves background elements such as music, ambient sound, and sound effects.
The model supports translation across more than 90 languages. To facilitate integration, an API for developers and large organizations was released on August 6, 2026. The original Dubbing v1 remains available for projects with budget constraints.
To promote the new technology, the company introduced the Creator Dubbing Partner Program. Participants receive up to 30 minutes of free usage during a seven-day promotional period.
