The patient storyteller
British · Warm · Unhurried
View description
An older British man with a low, lightly textured voice and clear pronunciation. He speaks at an unhurried pace, with the warmth of someone telling a story across a kitchen table.
AI Voice Design – Create Your Own Voice
Create a custom voice from a text description for videos, stories, or lessons. Test it with your own script, then save it to use again.
Your voice library
AI Voice Design Examples
Listen to four voices created from text descriptions. Choose a description you like and use it to create your own voice.
The practical guide
American · Clear · Conversational
View description
An American woman in her thirties with a clear, mid-range voice and a relaxed conversational delivery. Her tone is direct and friendly, with crisp consonants and little breathiness.
The weary gatekeeper
British · Gravelly · Restrained
View description
A middle-aged British man with a dry, gravelly voice and a restrained delivery. His words are distinct, and his baseline tone suggests someone who has answered the same question many times.
The approachable tutor
American · Gentle · Steady
View description
An American adult with a gentle, steady voice, a medium pitch, and precise pronunciation. The delivery is conversational and patient, with enough variation to keep an explanation easy to follow.
How to Write a Voice Description
Use these voice design tips to describe the speaker you have in mind and give the custom voice generator a clear starting point.
Instead of “a professional voice,” try:
A low-pitched voice with a slight rasp, a British accent, and a relaxed speaking pace.
Start with a few clear traits
A useful AI voice design prompt describes how the speaker sounds: approximate age, pitch, accent, and texture. Choose two or three traits that suit your project. Words like excellent or impressive give little direction.
Keep your voice design prompt consistent. Warm and reserved can work together; whispering and booming ask for different performances. Start with one or two sentences, then remove conflicting details. Longer descriptions are not always more useful.
Match the voice to your project
For a tutorial, test a number, a product name, and an instruction. For character voice design, compare ordinary dialogue with an emotional scene. For narration, listen against the footage or music you will actually use.
Use AI voice design to choose a speaker who suits your material. The same narrator should sound convincing in a welcoming introduction and a serious warning. Try both passages before saving, and keep the voice that works across your scripts.
Adjust one detail at a time
When a voice design preview misses the mark, identify the problem before changing the prompt. If the voice is too breathy, ask for less breathiness. If the accent is too strong, request a lighter one. Compare candidates with the same short passage to focus on the sound.
Listen for clear consonants, comfortable pauses, and pronunciation of names. Check a quiet sentence as well as an expressive one. Treat voice design as a short audition: choose the speaker who fits your material, then compare how that voice handles different lines.
AI Voice Design in Three Steps
Describe a speaker, hear a preview, and save the voice for future scripts.
Describe
Describe the speaker’s accent, tone, and pace, then generate a voice.
Listen
Listen to the preview, then open Try Text to Speech to hear the voice read your own script.
Save
Name and save the voice you like. Reuse it from your library with new text.
Voice Design vs. Voice Cloning
Voice design starts with a description. Voice cloning starts with a recording.
Create a New Voice
AI voice design creates a new speaker from a text description. Use this custom voice generator when preset voices don’t suit your project.
Design your voiceClone an Existing Voice
Recreate an existing voice from a recording. Use your own voice or one you have permission to clone.
Explore voice cloningAI Voice Design FAQs
Find answers about pricing, supported languages, saving voices, and using your audio.
AI voice design creates a custom voice from a text description. You describe how the speaker should sound, including their accent, pitch, and vocal texture. It can produce narrators, conversational speakers, or fictional characters without a reference recording.
Voice design creates a new speaker from a written description. Voice cloning uses a recording to reproduce an existing speaker's voice. Choose design when you are inventing a narrator or character; choose cloning for your own voice or one you have permission to clone.
AI voice design uses credits to create a voice, including its first preview. Listening to public examples and editing a description are free. Generating speech with your own text uses additional credits; saving a voice does not. The price is shown before each generation, and replaying an existing recording does not use credits.
A useful voice design prompt describes an accent, approximate age, pitch, vocal texture, and pace. Choose a few compatible traits, such as a clear mid-range voice with a light American accent and steady delivery. Avoid conflicting instructions, and change one detail when comparing previews.
The language menu includes English, Chinese, Japanese, French, Spanish, and more. Search for the language you want, then describe an accent in your voice design prompt. Test a short passage, as accent accuracy and pronunciation can vary.
Describe the overall tone and pace when creating your voice. Try Text to Speech uses that voice to read your text, with no separate emotion or speed controls. For a different sound, adjust your voice design description and create another preview. Each new voice uses credits; results can vary.
Yes. After creating a voice, open Try Text to Speech and enter up to 3,000 characters. Select Generate, then Download WAV to save the audio. The custom voice generator displays the speech cost before you submit. You can test your voice design before saving it. Download recordings you want to keep before the voice expires or you delete it.
Yes. Select Save voice, enter a name, and choose Save. Reuse it from your library or the text-to-speech editor. Unsaved voices expire one hour after the preview is ready; saved voices last up to one year from creation. Saving does not restart that period. Reusing your voice design avoids another creation charge, but new speech still uses credits.
A voice design prompt describes traits, not a fixed speaker. Each new creation can interpret those traits differently, even when the text is unchanged. To keep a particular result, save it and select that saved voice for new recordings. Reusing a voice preserves the selected speaker, though delivery can vary between readings.
Yes. You can use audio created with voice design in YouTube videos, courses, games, and other commercial projects under FlowSpeech's Terms of Service. You must have the rights to your script and any other material you use, and comply with applicable laws. Check your publishing platform's requirements before release.
Read the Terms of Service