Voice Design
Describe a voice in words, age, tone, accent, personality, and Kolbo generates a brand-new named custom voice
Type a description of any voice and Kolbo generates a brand-new voice from scratch. No recording, no sample audio, just words. The result is a named custom voice you can reuse across every project and character.
Overview
Voice Design creates a custom voice from a text description. It is the right choice when the voice you want lives only in your head and you have no sample to clone. Describe the age, tone, accent, emotional quality, and speaking pace, and Kolbo builds a unique voice that matches.
How It Works
-
Open Text-to-Speech in Audio Tools and open the voice creation panel
-
Choose "Design a voice"
-
Write a description of the voice you have in mind. A few sentences is enough:
"A warm, confident woman in her 40s. American Midwest accent. She speaks like a seasoned documentary narrator, unhurried, authoritative, with a slight smile in the voice."
-
Kolbo generates a unique voice from the description, names it, and saves it to your custom voice library
-
Select it for any Text-to-Speech run, assign it to a character, or reuse it across sessions and projects
What to Describe
The more specific the description, the better, though even a short one produces a distinct result:
- Age and gender - a teenage boy, a man in his late 60s, a young woman in her 20s
- Tone and mood - warm and encouraging, dry and skeptical, energetic and fast-paced, calm and meditative
- Accent or region - British RP, Southern US, Israeli, Brazilian Portuguese, Nigerian English
- Personality or archetype - a children's tutor, a tech CEO, a radio DJ from the 80s, a patient therapist
- Speaking style - measured with long pauses, quick and animated, deep and theatrical
Why It Changes the Workflow
Until now, getting a custom voice meant finding a recording, clipping a clean sample, and uploading it. That works well when the source exists, but most of the time the voice you want only exists in your imagination. Voice Design skips the recording entirely. Describe the character, get the voice. If the first result is close but not quite right, write a more specific description and try again. You can build an entire cast of named voices without recording a single second of audio.
Voices created through Voice Design live alongside cloned voices in the same library. They can be attached to Visual DNA characters, used in Text-to-Speech, and referenced from any project.
Tips
- Lead with the strongest attributes - age, accent, and archetype shape the result most
- Iterate with detail - if a take is close, add specifics rather than starting over
- Build a cast - design a named voice for each recurring character so it is ready everywhere