ElevenLabs Dubbing v2 preserves emotion tone pacing

ElevenLabs’ Dubbing v2 targets “emotion-preserving” translation

ElevenLabs has launched Dubbing v2, positioning it as a dubbing update meant to keep more than just words aligned when translating audio into many languages. The company’s central claim is that the voice output preserves the original speaker’s emotion, tone, and pacing while staying synchronized to the content.

That combination—prosody (tone and timing) plus synchronization—matters because typical text-to-speech or naive voice cloning approaches can produce audio that sounds fluent but emotionally “off,” or that drifts relative to on-screen timing. By emphasizing emotion, tone, and pacing together, ElevenLabs is targeting the common pain point that dubbed audio can feel mismatched to what viewers see and hear.

The rollout is also framed as multilingual scale: the product is described as supporting 90+ languages. The excerpt doesn’t specify whether those are all immediately available with the same quality tiers, or how the pipeline performs across different language pairs, but the focus is clearly on large coverage.

Why this matters is that dubbing is one of the fastest ways for streaming and media companies to expand global audiences without re-recording everything in each market. “Emotion-preserving” output is especially relevant for genres where delivery carries meaning—such as comedy, drama, and action-heavy dialogue—because viewers often notice timing and vocal nuance even when they can’t describe it.

From a developer and creator standpoint, the product is also an example of how voice AI vendors are trying to differentiate on higher-level constraints beyond just naturalness. Synchronization to content and retention of delivery characteristics are both engineering challenges.

As summarized here, the major announced change is the v2 model’s ability to keep emotional and rhythmic qualities aligned across a broad set of languages while producing dubbed audio that remains synced.