ElevenLabs Eleven v4: AI voices that act, not just read
ElevenLabs is the best-known name in AI voice. Its newest model, Eleven v4, reads a script the way a voice actor would, with real emotion, timing and character, in more than 90 languages.
Explore ElevenLabs ↗
- Made by
- ElevenLabs
- Model type
- AI voice model
- Released
- Eleven v4: 28 September 2026
- Best for
- Emotional, human-sounding voice-overs
What is ElevenLabs?
ElevenLabs released Eleven v4 on 28 September 2026, built on an entirely new design. It reads the whole script, works out who is speaking and what came before, and delivers each line with the right feeling, whether that's dramatic, tender, urgent or funny.
In ElevenLabs' blind listening tests, people preferred it in 65% to 81% of match-ups, and it ranked first on the Artificial Analysis voice leaderboard for September 2026. There is also Eleven v4 Turbo, a faster version built for live voice agents.
What ElevenLabs does best
- Real emotion. Lines sound acted, not read.
- Audio tags. Add [whispers], [laughs] or [excited] to direct the performance.
- Several speakers. Natural back-and-forth conversations in one take.
- 90+ languages. Up from over 70 in the previous version.
- Better voice cloning. Cloned voices sound closer to the original.
How to use ElevenLabs on Keoo AI
- Open Text to Speech in Keoo AI.
- Pick ElevenLabs from the model list.
- Paste your script, choose a voice, and add audio tags where you want emotion.
- Generate, listen, and download your voice-over.
What you can make with it
- Voice-overs for YouTube, TikTok and adverts
- Audiobooks and stories
- Podcast intros and dialogue scenes
- Course and training narration
- Character voices for games and animation
How to write a ElevenLabs prompt
- Put tags in square brackets right before the words: [whispers] I'll tell you a secret.
- Use punctuation for pauses. Eleven v4 has no speed control and doesn't read SSML break tags.
- For dialogue, give each speaker their own line and voice.
- Write how people talk: short sentences, contractions, natural rhythm.
Example 1: Two-voice scene (our sample)
Speaker 1: [excited] Have you heard it yet? The new trailer voice-over is done!
Speaker 2: [laughs softly] I have. Honestly? [whispers] It gave me goosebumps.
Example 2: Advert voice-over
[warm] Mornings are hard. [pause] Your coffee shouldn't be. [upbeat] Try Sunrise Roast today, and start every day on the right foot.
Questions people ask
What is the newest ElevenLabs model?
Eleven v4, released on 28 September 2026, with a faster Eleven v4 Turbo.
How many languages does Eleven v4 support?
More than 90.
What are audio tags?
Short directions in square brackets, like [whispers] or [laughs], that tell the voice how to perform a line.
Create more with ElevenLabs on Keoo AI
Keoo AI puts ElevenLabs next to the other top video, image and audio models in one dashboard. Open Music Generator, pick your model, and go from idea to finished content in minutes.
Start creating for free ↗