ElevenLabs launches Eleven v4
Eleven v4 is ElevenLabs' most expressive text-to-speech model, built for audiobooks, voiceovers, dubbing and character dialogue. A Turbo variant offers about 100 ms latency.
What it does
Eleven v4 is ElevenLabs' most expressive text-to-speech model, built on a new architecture to read tone, pacing, emotion and character from the text. You can describe delivery in plain language, or use inline tags such as [laughs] or [light rain] for emotion and sound effects.
Eleven v4 Turbo uses the same technology with a median latency of about 100 ms.
Results and limits
ElevenLabs says it's ranked #1 by Artificial Analysis and preferred by about 75% of listeners in blind head-to-head tests. Both figures come from ElevenLabs' own announcement.
Worth knowing
It's available in ElevenAgents, ElevenCreative and the ElevenAPI, and you can start with a free account. It supports IPA phonemes for exact pronunciation.
Try it yourself
[whispering, nervous] [Character name], did you hear that? [pause] [suddenly loud] Get down! [door slams]