You can start free: a new account comes with $2 of credit, and no card is needed. That reads about 97 minutes of speech with MAI-Voice-2.1, or 21 with MiniMax Speech 2.8 HD, at the pace of the samples on this page. After that you add credit when you want more and pay by the character. There is no subscription, and the welcome credit is one per person.

Which voice model should I use?

Eleven v3 is the one to try for acting: it follows tags in the script such as [whispers] or [laughs]. Gemini 3.8 Flash TTS takes a note on how to say the line, like the livestream host on this page. MiniMax Speech 2.8 HD and Eleven v4 suit voice-overs and ads, and MAI-Voice-2.1 and TTS-1 HD read long narration for less. The price of each sits next to its name, and you can read the same script with several and keep the take you like.

How long can the text be?

Up to 10,000 characters a run with MiniMax Speech 2.8 HD, and 4,000 with TTS-1 HD. The box on this page carries up to 280 characters through sign-in; paste a longer script into the studio. You pay for the characters you send.

Can I clone my own voice?

Not with these models: each one reads your words in its own voices, and none of them takes a recording to copy. The ElevenLabs Voice Changer can say a recording again in one of its preset voices. Upload only recordings you have the right to use, and never make a voice that passes for a real person who has not agreed: Leap's Acceptable Use Policy forbids impersonating anyone.

Can I use the voice-overs commercially?

Leap's Terms say that, as between you and Leap, you own the results the studio makes for you, to the extent the law allows, and they set no personal-use-only limit. In some countries AI-made content may not be protected by copyright.

The audio file the model returns, an MP3 or a WAV depending on the model. Each run stays in the studio with the words it was made from, so you can play it again or download it later.

What happens if a run fails?

It costs nothing. Only a run that succeeds is charged, and you see the price on the Generate button before you press it.

AI Voice Generator

Type a script, pick a model and a voice, and get a voice-over: Eleven v3, MiniMax Speech, Gemini and more text to speech models in one studio. Every voice on this page came out of Leap.

A kitchen out of limesEleven v3, voice Liam

0 / 9 s

Script

[excited] Table nine just ordered the entire left side of the menu! [shouting] Chef! We need more limes! [panting] All of the limes!

Use this voice

Heard on Leap, one script each

Real runs, as they came back, each with the script it was given and the model and voice that read it: narration, ads, podcasts, characters and a radio host at 3 a.m.

A fragrance ad, no product namedMiniMax Speech 2.8 HD, voice Elegant man
0 / 7 s
Seek in A fragrance ad, no product named
Script
It smells like the last train home, rain on warm stone, and a decision you will not regret in the morning.
Use this voice
A sneaker-drop livestream hostGemini 3.8 Flash TTS, voice Puck
0 / 8 s
Seek in A sneaker-drop livestream host
Script
Doors open at noon, sizes are limited, and no, your cousin cannot hold your place in line. I can see you, Kevin.
How to say it
The hyped host of a sneaker-drop livestream: fast, grinning, a little out of breath.
Use this voice
Documentary: seven miles downEleven v4, voice Lily
0 / 11 s
Seek in Documentary: seven miles down
Script
Seven miles below the surface, the pressure would fold a car like paper. And yet, down here, in total darkness, something is glowing.
Use this voice
A nervous first podcastEleven v3, voice Charlotte
0 / 13 s
Seek in A nervous first podcast
Script
[nervous] Hi, um, welcome to my podcast. This is episode one. [clears throat] Today we're talking about moths. [whispers] They're in the room with me right now.
Use this voice
Documentary: the high desert at nightMAI-Voice-2.1, voice Jasper
0 / 10 s
Seek in Documentary: the high desert at night
Script
In the high desert, the night drops forty degrees in an hour. The stones that baked all day begin to crack, and the whole valley ticks like a clock.
Use this voice
A maître d' with a warningTTS-1 HD, voice Sage
0 / 8 s
Seek in A maître d' with a warning
Script
Your table is ready. Please follow me, and mind the step. The chef has asked me to tell you that tonight, everything tastes better than it looks.
Use this voice
A sourdough confessionGrok TTS, voice Eve
0 / 10 s
Seek in A sourdough confession
Script
Okay, confession: I named my sourdough starter, and now I can't throw any of it away. It's been four years. It has a birthday. We sing.
Use this voice
A goblin potion sellerInworld TTS, voice Snik (en)
0 / 13 s
Seek in A goblin potion seller
Script
Buying or browsing? Browsing costs extra. Touching costs double. And if you breathe on the potions, we are going to have a very different conversation.
Use this voice
A 3 a.m. radio sign-offKokoro (American English), voice Af heart
0 / 12 s
Seek in A 3 a.m. radio sign-off
Script
It's three in the morning, in a city that pretends it sleeps. If you're still driving, crack a window. If you're still awake, text her back. This is Night Shift, and this next one goes out to the bakers.
Use this voice

Pick the model for the read

Each model has its own voices and its own way with a line. One account runs them all, so you can read a script with two and keep the better take.

Eleven v3$0.11 per 1,000 characters
ElevenLabs v3: expressive speech that follows audio tags like [laughs] or [whispers].
21 voices, up to 5,000 characters a run. By ElevenLabs.
Try Eleven v3
Eleven v4$0.088 per 1,000 characters
ElevenLabs' newest and most expressive speech model, in over 70 languages.
21 voices, up to 5,000 characters a run. By ElevenLabs.
Try Eleven v4
MiniMax Speech 2.8 HD$0.11 per 1,000 characters
MiniMax's studio-quality speech, with emotion and over 30 languages.
17 voices, up to 10,000 characters a run. By MiniMax.
Try MiniMax Speech 2.8 HD
Gemini 3.8 Flash TTS$0.0495 per 1,000 characters
Google's speech model: describe the delivery in style instructions.
30 voices, up to 4,000 characters a run. By Google.
Try Gemini 3.8 Flash TTS
MAI-Voice-2.1$0.0242 per 1,000 characters
Microsoft's most natural voices, for narration, across 23 languages.
3 voices, up to 4,000 characters a run. By Microsoft.
Try MAI-Voice-2.1
TTS-1 HD$0.033 per 1,000 characters
OpenAI's higher-quality text to speech, in nine voices.
9 voices, up to 4,000 characters a run. By OpenAI.
Try TTS-1 HD

See every voice and sound model

What it costs

Start free, then pay by the character

A new account comes with $2 of credit, and no card is needed. Speech is billed by the character, so the table shows each model's price per 1,000 characters and about how much speech $2 reads. A run that fails costs nothing.

Model	Price	What $2 makes
Eleven v3	$0.11per 1,000 characters	About 21 minutes18,181 characters
Eleven v4	$0.088per 1,000 characters	About 26 minutes22,727 characters
MiniMax Speech 2.8 HD	$0.11per 1,000 characters	About 21 minutes18,181 characters
Gemini 3.8 Flash TTS	$0.0495per 1,000 characters	About 47 minutes40,404 characters
MAI-Voice-2.1	$0.0242per 1,000 characters	About 97 minutes82,644 characters
TTS-1 HD	$0.033per 1,000 characters	About 71 minutes60,606 characters

Prices are the live catalog's, per character as the models bill them. Minutes are at 14 characters a second, the pace of the voices on this page. How pricing works.

Questions

Questions people ask

Is it free?: You can start free: a new account comes with $2 of credit, and no card is needed. That reads about 97 minutes of speech with MAI-Voice-2.1, or 21 with MiniMax Speech 2.8 HD, at the pace of the samples on this page. After that you add credit when you want more and pay by the character. There is no subscription, and the welcome credit is one per person.
Which voice model should I use?: Eleven v3 is the one to try for acting: it follows tags in the script such as [whispers] or [laughs]. Gemini 3.8 Flash TTS takes a note on how to say the line, like the livestream host on this page. MiniMax Speech 2.8 HD and Eleven v4 suit voice-overs and ads, and MAI-Voice-2.1 and TTS-1 HD read long narration for less. The price of each sits next to its name, and you can read the same script with several and keep the take you like.
How long can the text be?: Up to 10,000 characters a run with MiniMax Speech 2.8 HD, and 4,000 with TTS-1 HD. The box on this page carries up to 280 characters through sign-in; paste a longer script into the studio. You pay for the characters you send.
Can I clone my own voice?: Not with these models: each one reads your words in its own voices, and none of them takes a recording to copy. The ElevenLabs Voice Changer can say a recording again in one of its preset voices. Upload only recordings you have the right to use, and never make a voice that passes for a real person who has not agreed: Leap's Acceptable Use Policy forbids impersonating anyone.
Can I use the voice-overs commercially?: Leap's Terms say that, as between you and Leap, you own the results the studio makes for you, to the extent the law allows, and they set no personal-use-only limit. In some countries AI-made content may not be protected by copyright.
What do I download?: The audio file the model returns, an MP3 or a WAV depending on the model. Each run stays in the studio with the words it was made from, so you can play it again or download it later.
What happens if a run fails?: It costs nothing. Only a run that succeeds is charged, and you see the price on the Generate button before you press it.

Your script, read back in seconds.

Start free, no card needed. Every run is priced before you press Generate.

Make a voice-over

Heard on Leap, one script each

Pick the model for the read

Start free, then pay by the character

Add music and video

Questions people ask

Your script, read back in seconds.