A Voice Design prompt can be simple. A short description such as “a young woman with a light French accent” is enough to start generating a voice.
For a more specific result, add the details that matter to you: role, audible traits, delivery, or emotional context. You do not need every element, and more detail does not guarantee a better voice. A matching preview script helps you hear how the result fits its intended use.
Five details you can add to a Voice Design prompt
Depending on the voice you want, you can add any of the following:
Role and situation
Base voice
Vocal texture
Pace and delivery
Emotion and script fit
1. Establish the role and situation
Begin by telling us who the speaker is and where we would naturally hear them.
Useful roles include:
A customer support agent near the end of a long shift
A late-night radio host speaking to someone who cannot sleep
Compare:
A warm male voice.
With:
Create the impression of a late-night radio host speaking to someone who cannot sleep.
The second version immediately suggests intimacy, volume, pacing, and the relationship with the listener.
2. Define the base voice
Describe the voice's approximate age, gender when relevant, pitch, accent, and natural or synthetic character using everyday language. For example:
a young woman with a light French accenta middle-aged man with a deep, warm voicea friendly digital assistant with a light synthetic quality
3. Use audible texture words
Texture describes the surface of a voice. Useful words include breathy, raspy, gravelly, velvety, crystalline, nasal, smoky, polished, weathered, and glossy.
Use contrasts to make decisions. Is the voice polished or weathered? Airy or grounded? Crystalline or smoky? One or two clear texture choices usually work better than a long stack of near-synonyms.
4. Describe pace and delivery
Voice identity includes habitual movement through language. Describe speed, articulation, pauses, and phrasing:
slow and deliberaterapid-firecrisp articulationheavy dramatic pauses
Describe the voice's usual delivery here. Use Voice Direction later to adjust a saved voice for individual scenes—for example, from slow narration to urgent dialogue.
5. Define emotion and script fit
Finish by explaining what kinds of emotional material the voice should carry well:
Urgent without becoming panicked
Grounding, tranquil, intimate, and safe
Tired but capable and professional
Threatening through composure rather than volume
Warm and helpful with subtle robotic consistency
This defines the voice's natural center and the kind of material used to audition it. It does not mean every future line must use the same emotion.
Combine your chosen details into a few natural sentences. Explain how the traits work together, such as a tired support agent who remains patient and professional. The examples below show how these choices can form a short voice description.
Use a role-matched preview script when it helps
A preview script that the character would naturally say makes it easier to hear the qualities you described.
For the tired support agent:
I understand why that charge is frustrating, and I really do want to help. Let me pull up the account one more time, but I need to be honest with you: this particular fee is marked as non-refundable.
This gives you a chance to hear fatigue, professional restraint, and empathy in the same passage.
Four expanded prompt-and-script examples
Meditation guide
Voice description
A calm female meditation guide with a soft, breathy voice. Speak slowly and smoothly, with gentle pauses and a warm, reassuring tone that feels peaceful, intimate, and safe.
Preview script
Let your shoulders drop away from your ears, releasing any tension you might be holding. Breathe in deeply through your nose, feeling the air expand your chest, and then slowly exhale, letting everything soften into this present moment.
Listen: meditation guide
Elegant villain
Voice description
A middle-aged man with a deep, smooth voice, playing an elegant villain. Speak slowly and clearly with controlled menace; keep the tone calm, intimate, and faintly amused, never loud.
Preview script
Did you truly believe you could dismantle my life's work without consequence? I admire the audacity, really, I do. But now that the doors are locked, let us discuss exactly how much your ambition is going to cost you.
Listen: elegant villain
Older chronicler
Voice description
An older man with a deep, slightly rough and weathered voice. Speak slowly with measured pauses and restrained emotion, giving narration, warnings, and dark memories a quiet sense of weight.
Preview script
People think ruin announces itself with thunder. It does not. It sits beside you quietly, waits for compromise, and learns your name before you notice the door is locked.
Listen: older chronicler
Polished digital assistant
Voice description
A friendly digital assistant with a bright, clear voice and a subtle synthetic quality. Speak at a crisp, steady pace with precise pronunciation, sounding warm, helpful, and consistently composed.
Preview script
Thank you for waiting. I have located your order details and will now summarize the current status, estimated delivery window, and available support options.
Listen: polished digital assistant
Choose the right guidance setting for Voice Design
Guidance controls how strongly a generation interprets the Voice Design prompt. The API and web product expose this control differently.
In the API
guidance_scale accepts integer values from 1 to 10. To balance creativity and stability in Voice Design, set it between 2 and 4. Pass an explicit value when you want controlled comparisons between prompt revisions.
Stay nearer the lower end when consistency matters most.
Move upward when the design needs stronger character or emotional expression.
Change the value gradually and keep the prompt and script unchanged while comparing.
In the web product
The web product offers Stable, Expressive, Creative, and Intense. Start with Stable for controlled delivery or Creative for more character and expression. These are our recommended starting choices. Explore the other levels as needed, keeping the description and preview script fixed when comparing.
An optional drafting template
Use the lines that help describe your idea:
[Name or role] sounds like [speaker in a concrete situation].
They speak in [age/gender/register/accent or synthetic quality].
The voice has [one or two audible textures].
The delivery is [pace, articulation, pauses, rhythm].
It should fit [emotional range and kinds of lines], with [one defining boundary or contrast].
Preview script:
[One or two role-specific sentences that reveal those traits.]Listen to the preview, then refine one detail at a time to move closer to the voice you imagine.