The ElevenLabs library
Filter by language, accent, gender, age and use case, and audition any voice before you use it.
VoicesElevenLabs
Text to speech. Any script, the voice it needs.
Paste the script, pick a voice from the ElevenLabs library or a preset, and get spoken audio on your board. The engine, the emotion and the speed sit in the same dock.
Generate speech
Free to start. No credit card.
- SCRIPT
- 5,000
- characters per run
- ENGINES
- ElevenLabs
- v3 · Multilingual v2 · Flash v2.5
- SPEED
- 0.5–2x
- 1x by default
- OUTPUT
- MP3
- 32 to 192 kbps
Every voice-over, one dock.
The model, the voice and the delivery are set where you write the script.
Delivery in the script
v3 reads emotion tags, and Prompt Emotion rewrites the script for the feeling you pick.
Pace and format
Set the speed from 0.5x to 2x and take the file as MP3, from 32 to 192 kbps.
Hear it before you press.
Every voice plays in a click, from the ElevenLabs presets to MiniMax’s ten, so the voice you pick is the one you heard.
Browse voicesGeorge
ElevenLabs
Aria
ElevenLabs
Brian
ElevenLabs
Lily
ElevenLabs
Calm Woman
MiniMax Speech 2.8 HD
Deep Voice Man
MiniMax Speech 2.8 HD
Write the delivery into the script.
Tag a line [excited] or [whispers] and v3 reads it that way. Prompt Emotion rewrites the whole script for one feeling, and Undo puts it back.
Generate speech
[excited] It’s finally here. [whispers] And only until Sunday.[excited] It’s finally here. [whispers] And only until Sunday.
Generate
From voice-over to video.
Send the take to the Video Editor to lay it under a cut, or upload it in UGC Studio and a creator lip-syncs to it.
Generate speechHow to turn a script into a voiceover
Four steps in Voice Studio.
Introducing Text to Speech
Generate speechScripts to spoken audio in the voice the brief calls for: the ElevenLabs library and presets, three ElevenLabs engines and MiniMax, with emotion, speed and format in the same dock.
FAQs
A Voice Studio job that turns a script into spoken audio. Pick a model and a voice, write the script and press Generate.
The ElevenLabs library, with filters for language, accent, gender, age and use case; the ElevenLabs presets; and the ten preset voices of MiniMax Speech 2.8 HD.
Up to 5,000 characters per run on ElevenLabs and MiniMax.
Multilingual v2 is the most stable across 29 languages. MiniMax Speech 2.8 HD takes 17 languages, or Auto.
Yes. v3 reads emotion tags such as [excited] and [whispers], MiniMax has its own emotion list, and on models that offer it the speed runs from 0.5x to 2x.
The credits for the run are on the Generate button before you press it. Flash v2.5 costs half as much as v3 and Multilingual v2.
Yes. Attach a voice sample on a clone-capable model, such as Koi Voice Clone.