Sparkles EditorImagination, made real.Credits & billing
All tools

AI voice design: describe a voice with text

Voice design lets you describe vocal qualities in words and have that voice read a script. It is useful for exploring an original narrator or fictional character. Unlike voice cloning, it does not require a reference recording to get started.

Open tool

Updated

What you provide
A required voice description and an English or Chinese script, each up to 2,000 characters.
What you get
Private WAV speech guided by your description, with playback, script reuse, download and deletion.
Reference audio
No recording required
Required fields
Voice description + script
Main languages
English and Chinese
Cost and retention
Currently free for a limited time · WAV kept for 7 days

What you can make

An original character voice

Describe a fictional character’s manner, tone and rhythm to explore narration for a story or storyboard.

Different brand delivery styles

Compare calm, bright or friendly approaches using the same short passage before choosing a direction.

Documentary and story reading

Write down emotion and pause requirements so the delivery suits the material.

How to use it

  1. Open the studio and choose Voice design. Prepare a short English or Chinese passage to use for comparison.
  2. In Style instructions, describe the role, vocal texture, pitch, pace and emotion. Start with two or three priorities and avoid conflicting directions.
  3. Put only the words to be read in Script. Script and description each allow up to 2,000 characters; the description is required in this mode.
  4. Generate and listen for clarity, rhythm and emotion. Change only one or two details at a time and keep the test script the same.
  5. Download the version that fits. Reuse the record to revise the script and generate again; the voice may still vary between requests.

Start with an example

Design a gentle storyteller

Describe general vocal qualities for an original character without naming a real person’s identity.

Voice description
A warm, clear adult narrator with a medium-low register and an unhurried pace. Gently curious, like telling a quiet bedtime story. Restrained emotion and natural pauses between sentences.
Script
The little lighthouse began to glow. A breeze passed the window, bringing a story from far away into the stillness of the night.

Check the result: If it is too slow, change the pace first. If it is too expressive, simplify the emotion words. Small changes make comparisons easier.

Open tool

Common questions

Should I choose voice design or voice cloning?

Choose design to explore an original voice without a recording. Choose cloning to reference your own or an authorized voice. For a quick reading, start with a preset.

How can I make a voice prompt easier to refine?

Describe role, texture, pace and emotion separately, prioritizing two or three qualities. Keep the test passage unchanged and adjust one or two instructions per comparison.

Does this create music or singing?

This page describes speech synthesis and narration, not music or song production. It also does not assemble the audio into a finished video.