AI voice generator
Turn written text into speech with 67 ElevenLabs voices. The AI voice generator offers five models, Eleven v4, Eleven v4 Turbo, Eleven v3, Turbo v2.5 and Multilingual v2, takes up to 5,000 characters in one run and prices the run by the characters you type, so the cost line under the box is the price you pay. Preview any voice before you spend a credit, adjust the voice settings each model takes, and play or download the result.
The AI voice generator in numbers
ElevenLabs Eleven v4, Eleven v4 Turbo, Eleven v3, Turbo v2.5 and Multilingual v2, billed at 10, 5, 12.5, 6.25 and 12.5 credits for every 1,000 characters. The total is rounded up once, so a short line costs a credit, not a full block.
Premade ElevenLabs voices, each described in the provider's own words, such as Deep, Calm or Narrator. Search by name, narrow by those words, and play a free preview of any voice.
The most one run takes, on all five models. Spaces and punctuation count; spaces at the start and end are trimmed before the run is priced.
From 1 credit for up to 200 characters on Eleven v4 Turbo to 63 for 5,000 characters on Multilingual v2 or Eleven v3. The cost line counts your characters as you type.
Stability, similarity and style from 0 to 1, and speed from 0.7 to 1.2, under Advanced. Turbo v2.5 and Multilingual v2 take all four, Eleven v4 and v4 Turbo take stability and similarity, and Eleven v3 takes stability only. They change how the voice sounds and never the price.
Every model except Multilingual v2 takes a language setting with 31 choices and Auto-detect as the default. Multilingual v2 speaks 29 languages and finds the language from the text; Eleven v3 speaks 70+ languages, and Eleven v4 and v4 Turbo 90+.
Watch the AI voice generator set up a line
This page recorded at about 1.25 times speed, without sound.
An example, a voice, a sample
Game character line fills the text box, the voice picker opens on its search and style filters, and Product intro swaps in a second example, priced at 1 credit for its 109 characters. The recording ends on the second sample line in the player.
Choosing a voice
Find the voice before you spend a credit
The voice list is the provider's catalog, shown the way the provider describes it.
Free previews
Every voice has a short sample clip you can play from the list. A preview runs no model and takes no credits. It is the provider's sample line, not your text.
Search and filter
Type a name or a word from a description, or tap a filter such as Deep, Calm, Energetic, Narration, Announcer or Character. A filter appears only where two or more voices share the word.
Descriptions as published
The provider gives each voice a few words, for example Husky, Engaging and Bold, or Calm, Soothing and Meditative. It publishes no gender, age or accent field, so the list has none; British and Australian filters exist only where a description says so.
Premade voices only
The generator does not clone a voice from a recording and does not offer voices of real people. Pick from the 67.
Eleven v4, v4 Turbo, v3, Turbo v2.5 or Multilingual v2
All five read the same text with the same 67 voices; they differ in price, language control and the voice settings they take. A 400-character line costs 2 credits on Eleven v4 Turbo, 3 on Turbo v2.5, 4 on Eleven v4 and 5 on Multilingual v2 or Eleven v3.
| Model | Credits | Languages | Voice settings |
|---|---|---|---|
| Eleven v4Most emotive | 10 per 1,000: 1,000 characters cost 10 and 5,000 cost 50 | 90+ spoken; 31 to pick, or Auto | Stability and similarity |
| Eleven v4 TurboReal time | 5 per 1,000, the lowest: 1,000 characters cost 5 and 5,000 cost 25 | 90+ spoken; 31 to pick, or Auto | Stability and similarity |
| Eleven v3Expressive | 12.5 per 1,000: 1,000 characters cost 13 and 5,000 cost 63 | 70+ spoken; 31 to pick, or Auto | Stability only |
| Turbo v2.5All four settings | 6.25 per 1,000: 1,000 characters cost 7 and 5,000 cost 32 | 31 to pick, or Auto | Stability, similarity, style and speed |
| Multilingual v2Reads the language from the text | 12.5 per 1,000: 1,000 characters cost 13 and 5,000 cost 63 | 29, with no language setting | Stability, similarity, style and speed |
From text to speech in three steps
Text, voice, run.
- 1
Type or paste up to 5,000 characters, or tap an example line to fill the box. The cost line shows how many characters you have and what they cost.

- 2
Pick a model, search the voices and play a few previews, then set a language on any model except Multilingual v2, or leave Auto-detect.
- 3
Run it for the price on the button. The result appears in the results panel with a player and a download.

Limits of the AI voice generator
What the generator takes in, what it does not offer, and where a result can differ from what you hoped for.
5,000 characters a run
Longer scripts need to be split into several runs. The price follows the characters, so a split costs about the same as one long run.
No cloning and no real people
Only the 67 premade voices are available. The generator cannot imitate a person from a sample.
Five ElevenLabs models
The generator serves Eleven v4, Eleven v4 Turbo, Eleven v3, Turbo v2.5 and Multilingual v2. Other ElevenLabs models, such as Flash v2.5, are not offered here.
Language control is not on every model
Multilingual v2 refuses a language setting, so it reads whatever language the text is in. The other four list 31 languages in the setting; Filipino is not among them, because the provider asks for two-letter codes and Filipino has none. Eleven v3 speaks 70+ languages and Eleven v4 90+, so leave Auto-detect for one the list does not name.
Voice settings are a nudge
Stability, similarity and style shape the delivery without promising a particular result, so the same text can sound different from one run to the next. Eleven v4 and v4 Turbo take stability and similarity; Eleven v3 takes stability only. Re-run for another take; each run is billed.
Say it is synthetic
The audio is generated. Tell your audience that a voice is AI-generated wherever they would expect to be told, such as a published video or a game trailer.
Paid per character
You pay for each run by its characters, quoted in the cost line before it starts. A run that fails gives its credits back.
Where the voice goes next
Make the picture, the clip or the character the line belongs to.
Make a clip to talk over
The AI video maker makes a short video from a picture or a prompt; lay the generated voice over it in your editor.
Animate a 3D model
3D model to video turns a render of your model into a short clip, ready for a voiceover.
Show it turning, free
The 3D turntable video renders your real mesh turning 360° in your browser, a quiet shot for narration.
Questions about the AI voice generator
Models, voices, languages, length and cost.
Hear your first line
Type a line, play a few voice previews, and run it for the price in the cost line.