Krux
March 11, 2026
Voice AI Now Clones Actors From One Second of Audio
Published: March 11, 2026 at 12:30 AM
Updated: March 11, 2026 at 12:30 AM
100-word summary
Deepdub's new Phantom X 3.2 model can clone a voice from just one second of reference audio, then dub that performance across 20 languages simultaneously. The model layers emotions like joy and laughter into single lines and locks in consistent pronunciation for character names across entire seasons. It also powers real-time voice agents with 125-millisecond latency, fast enough for natural back-and-forth conversations. The catch: it's enterprise-only, available through Deepdub's GO platform. Netflix-scale localization just got cheaper, but the release sidesteps thorny questions about consent and who owns a cloned voice.
What happened
Deepdub's new Phantom X 3.2 model can clone a voice from just one second of reference audio, then dub that performance across 20 languages simultaneously. The model layers emotions like joy and laughter into single lines and locks in consistent pronunciation for character names across entire seasons. It also powers real-time voice agents with 125-millisecond latency, fast enough for natural back-and-forth conversations. The catch: it's enterprise-only, available through Deepdub's GO platform.
Why it matters
Netflix-scale localization just got cheaper, but the release sidesteps thorny questions about consent and who owns a cloned voice.