ElevenLabs ships Eleven v4: audio tags direct emotion and voice clones are indistinguishable

thealexbanks · x · 2026-09-29

ElevenLabs has released Eleven v4, its most expressive voice model yet — the author says he can no longer tell his cloned voice from the real one.

Key features:

Real-world use cases include audio newsletter overviews in your own cloned voice, video ad hooks from proven email lines, one voice across all dubbing languages, and fixing a single audiobook sentence without re-recording the chapter. Workflow: pick Eleven v4 in Text to Speech, choose a voice, and direct delivery via tags.

Related event: ElevenLabs Launches Eleven v4, Tops TTS Leaderboards at a Premium Price(22 posts)→

Original post →

More from Multimodal

Multimodal channel →