How to design a custom AI voice
- Describe the voiceIn the first box, write who is speaking: age, gender, tone, accent and mood. One or two descriptive sentences work best.
- Write what it should sayIn the second box, type the script for that voice. Keep it in the same language you select below.
- Choose the languagePick English, Spanish, French, German, Italian, Portuguese, Russian, Chinese, Japanese or Korean.
- GeneratePress Design & Generate. Your new voice appears on the right.
- Refine and downloadNot quite right? Change one word in the description (for example “raspy” to “smooth”) and generate again. Download the take you like.
How to write a voice description that works
The AI voice designer listens to the words you choose, so be specific. Think about the voice the way a casting director would. A strong description covers five things:
- Age: “a teenager,” “a woman in her 30s,” “an elderly man.”
- Gender and pitch: “a low male voice,” “a bright, high female voice.”
- Tone and texture: warm, crisp, gravelly, breathy, smooth, nasal.
- Accent: “General American,” “a light Texas accent,” “British RP,” “Irish.”
- Mood and pace: calm and slow, excited and fast, serious, playful, sarcastic.
Weak: “A nice voice.”
Strong: “A confident woman in her 40s, General American accent, warm and clear, speaking at a steady pace like a morning news anchor.”
Avoid naming real people (“sounds like a famous actor”). Describe the qualities you like instead; it is both more effective and the right thing to do.
Ready-to-use voice description ideas
Copy one of these into the description box and adjust it to taste.
- Audiobook narrator: “A calm man in his 50s, rich and low voice, neutral American accent, unhurried storytelling pace.”
- Kids’ cartoon hero: “An energetic young boy, high and bright voice, cheerful and a little mischievous, speaking quickly.”
- Fantasy game villain: “An old man with a deep, gravelly, hushed voice, slow and menacing, every word deliberate.”
- Meditation guide: “A soft-spoken woman in her 30s, gentle and breathy, very slow and soothing.” Pairs well with our meditation voice tips.
- Brand spokesperson: “A friendly, upbeat woman in her late 20s, crisp American accent, smiling as she talks.”
- Trailer announcer: “A powerful, booming male voice, dramatic pauses, intense and cinematic.” See also movie trailer voice.
Why use an AI voice designer
A voice nobody else has
Stock voices are shared by thousands of videos. A designed voice gives your channel or brand its own sound.
No recording required
Unlike voice cloning, you do not need a sample, a quiet room or permission from a speaker.
Perfect for characters
Build a full cast for an indie game, audio drama or animated short, one description per character.
10 languages
Design voices in English, Spanish, French, German, Italian, Portuguese, Russian, Chinese, Japanese and Korean.
Commercial use included
Use designed voices in YouTube videos, ads, apps and courses.
Your own account
Open the page and start designing. There is nothing to sign up for.
Building a cast for a podcast drama or game
Planning several characters? Keep a simple “voice bible” in a doc or spreadsheet. For each character, save the exact description you used, the language, and the file name of the take you approved. When you need new lines later, paste the same description, generate a few takes and pick the closest match.
Contrast is your friend. Give each character a clearly different age, pitch and pace, for example a slow, deep sheriff, a fast-talking teenage deputy and a dry, sarcastic narrator. Listeners can then follow a scene without being told who is speaking. Writing for younger audiences? Our kids’ story narration page has more ideas.
Voice Design at a glance
| Feature | On Speechma Pro |
|---|---|
| Input | A written voice description plus your script |
| Script length | up to 10,000 characters per conversion generation |
| Languages | 10 (English, Spanish, French, German, Italian, Portuguese, Russian, Chinese, Japanese, Korean) |
| Output | Audio you can play and download |
| Sign-up | Yes, a Speechma Pro account |
| Commercial use | Included |
Each generation creates a fresh voice from your description, so save the takes you like. For hundreds of ready-made voices with 10,000 characters per conversion, use text to speech.
Voice Design in 10 languages
Each language has its own guide with tips for that language:
English · Spanish · French · German · Chinese · Japanese · Korean · Russian · Portuguese · Italian
Frequently asked questions
What is AI voice design?
Is the AI voice designer included in Speechma Pro?
Will I get the exact same voice every time?
What is the difference between voice design and voice cloning?
Can I design a voice that sounds like a celebrity?
Can I use a designed voice in a commercial project?
Make better voiceovers, faster
- New voices and features the day they launch
- Script templates and tricks that make AI voices sound human
- Creator stories, community picks and early access