ElevenLabs AI Voice Generator (ElevenLabs Multilingual v2, v3 and Flash v2.5)
Natural AI voiceovers from a script, in three tiers: Natural, Expressive and a fast draft read.
ElevenLabs makes some of the most widely used text to speech models in the world, and this page covers the three that run the AI Voice Generator here. You write the script, pick a voice from the library or design a new one from a description, and get back a voiceover with captions timed to every word.
The tiers are different jobs. Natural (Multilingual v2) is the steadiest, most natural read, in 29 languages, and the right default for ads, explainers and narration. Expressive (v3) has more emotion and range across more than 70 languages, and follows direction tags written into the script, such as [excited], [whispers] or [laughs]. Fast draft (Flash v2.5) is quick and costs less per character, for timing a script before the final read.
There is no voice cloning here. You choose from the library, or describe a voice (age, accent, tone, pace) and get three designed versions to keep, which also keeps an ad clear of impersonation problems. Starting from a voiceover you already like works the same way: Audio to Prompt describes the voice and the delivery without identifying anyone, and the generator finds the closest library voices or designs one from that description.
What ElevenLabs Voices is best at
Ad and UGC voiceovers
Explainers and demos with timed captions
Expressive reads directed with tags
Scripts in 29 to 70+ languages
Natural by default The Natural tier gives a steady, believable read that suits ads, explainers and long narration without drifting in tone.
Direction in the script On the Expressive tier, tags like [excited], [whispers] or [laughs] change the delivery at that point in the script, so one read can shift mood.
Captions included Every voiceover comes back with SRT and VTT captions timed to each word, ready for a video editor.
Voices without cloning Pick from the library or design a voice from a description. Nobody's real voice is copied, so the read is yours to use.
Where ElevenLabs Voices falls short
No voice cloning. You cannot upload a recording and get that person's voice; describe the voice instead and design one, or pick the closest library match.
Scripts that imitate a real person, put words in a politician's mouth or read like a call from a bank or a government body are refused before anything is charged.
Each read takes up to 5,000 characters. Longer scripts go in parts.
Direction tags are an Expressive feature. The generator offers them only on that tier, so keep them out of Natural and Fast draft scripts.
ElevenLabs Voices specifications
Provider: ElevenLabs
Model ID: eleven_multilingual_v2, eleven_v3, eleven_flash_v2_5
Tiers: Natural, Expressive, Fast draft
Languages: 29 on Natural, 70+ on Expressive
Length: Up to 5,000 characters a read
Best at: Voiceovers with timed captions
Provider model ID: eleven_multilingual_v2, eleven_v3, eleven_flash_v2_5
What you can do with ElevenLabs Voices
Ad voiceover A hook and a call to action in a voice that fits the brand. The script goes over; pick the voice in the generator.
Explainer narration Clear, steady narration for a walkthrough or a product demo, with captions timed to every word.
Expressive read More emotion and range on the Expressive tier. Write direction tags where the delivery should change, then pick Expressive in the generator.
Design a new voice Describe the voice you want (age, accent, tone, pace) and get three designed versions to choose from and keep.
Make it with AI Voice Generator, which runs ElevenLabs Voices and shows the price before you spend anything. To start from a track, a voiceover or a sound that already works, read it with Audio to Prompt first: it describes the sound without naming anyone and hands that description to the generator.
Prompts that work on ElevenLabs Voices
Write for the ear
Short sentences. One idea each. Spell out numbers the way they should be said: twenty percent, not 20%.
A voice reads what is written. Short sentences give it places to breathe, and spelled-out numbers stop it guessing how to say them.
Direct the Expressive tier
[excited] It is finally here! [whispers] And it costs less than you think. [laughs] Okay, go and see for yourself.
On Expressive, a tag changes the delivery from that point on. One or two per script works; a tag on every line flattens back into noise.
Start from a voiceover that works
Upload an ad read you like to Audio to Prompt, then send its voice description to the generator to find the closest library voices or design a new one.
Audio to Prompt describes age, accent, pace and delivery without identifying the speaker, which is how you get the feel of a read without cloning anyone.
What ElevenLabs Voices costs
Cost on Prompt Reverse, per 100 characters, natural: 1 AI Credit, per 100 characters, expressive: 1 AI Credit, per 200 characters, fast draft: 1 AI Credit, a 30-second ad read, natural: 4 AI Credits, designing a new voice: 5 AI Credits. AI Credits come from a one-time credit pack and never expire, with no subscription and no separate provider account.
ElevenLabs Voices vs the other models
ElevenLabs Sound Effects: Switch for sounds rather than speech: whooshes, clicks, foley and ambience.
Lyria 3.5: Switch when the words should be sung, as a jingle or a song.
Stable Audio 3: Add a bed under the voiceover, cut to the same length as the edit.
Three: Multilingual v2 as the Natural tier, v3 as the Expressive tier and Flash v2.5 as the Fast draft tier. All three run in the AI Voice Generator from the same credit balance.
Can I clone a voice?
No. You can pick a voice from the library or design a new one by describing it. Real people's voices are not copied, which also keeps an ad clear of impersonation issues.
What languages does it speak?
The Natural tier reads 29 languages and the Expressive tier more than 70. Write the script in the language you want and pick a voice that suits it.
Can I use the voiceover commercially?
Yes, in ads, videos, podcasts and social content, online and offline. Platforms like YouTube ask you to label realistic AI voices, and every file is tagged as AI-generated in its metadata to make that simple.
Do I get captions?
Yes. Every voiceover comes with SRT and VTT captions timed to each word, ready for your video editor.
How is a voiceover priced?
By the length of the script, per block of characters, and the Fast draft tier costs less per character. The exact price is shown before you run it.