Describe a voice. Get three. Keep the one you want.
Voice Design creates a voice that has never existed from a written description — no recording, no sample, and nobody's permission to obtain. Designing and comparing previews is free; a voice slot is used only when you save one.
How Voice Design works
Three steps, and the first two cost nothing.
- 1
Describe it
Write 30–250 characters covering age, gender, accent, pitch, pace and tone. The description itself is written in English — it's the brief, not the script.
- 2
Compare three previews
Each attempt returns three distinct voices speaking your preview line. Not right? Try again — up to three attempts per session, all free.
- 3
Publish the keeper
Name it, tag it, and it joins your voice library — ready to use in Studio and Voice Studio like any other voice.
Why design instead of clone
Cloning copies a real person. Designing invents someone who doesn't exist.
No audio needed
There's nothing to record and nothing to upload. A sentence of description is the whole input.
No consent to chase
A designed voice belongs to no one, so there's no speaker to get permission from and no likeness to worry about.
Free to design and compare
Generating previews costs nothing. A voice slot is only used at the moment you save a voice to your library.
Three at a time
Every attempt returns three options, so you're choosing between voices rather than accepting the first one.
16 preview languages, Turkish included
Preview your design in Turkish, English, German, Spanish and 12 more — and a published voice keeps the same identity in every language it speaks.
Yours to keep and reuse
A designed voice sits in your library permanently and can be cast in any Studio or Voice Studio project.
When designing beats every other option
Cases where there is no real voice to start from.
A brand voice
One consistent voice across every video, ad and IVR prompt — that doesn't resign, raise rates, or become unavailable.
Characters
A cast for a game, animation or audio drama, where each character needs a voice nobody will recognise.
Narration with a specific feel
When you know exactly how the narrator should sound and no stock voice quite gets there.
When cloning isn't appropriate
No willing speaker, no usable recording, or simply a project where copying a real person is the wrong call.
Frequently asked questions
?Do I need to record anything?
No. Voice Design takes a written description only. If you want to reproduce a specific real voice, that's Voice Cloning instead.
?Why does the description have to be in English?
The design model reads the brief in English. It only describes the voice — what the voice then says is a separate field, and the published voice is not limited to English.
?Can I design a voice that speaks Turkish?
Yes. Turkish is one of the preview languages, so you can hear each option speaking Turkish before you publish, and the published voice narrates Turkish in Studio and Voice Studio.
?What makes a good description?
Concrete attributes beat adjectives. Age range, gender, accent, pitch, speaking pace and tone in one or two sentences will get you closer than a long, abstract description.
?What does it cost?
Designing and comparing previews is free. Saving a voice uses one of your plan's voice slots — see pricing for how many each plan includes.
?What if none of the three are right?
Adjust the description and try again — three attempts per session, each returning three fresh options. Previews you don't publish are cleaned up automatically.
?Where can I use the voice afterwards?
Anywhere a library voice works: single narration in Studio, or cast as a speaker in a multi-speaker Voice Studio project.
The voice in your head doesn't exist yet
Describe it and hear three versions of it. Designing costs nothing.