Steps
Describe the voice you want
Use You get up to 3 voice options. Each includes a
voice create with a natural language prompt:preview_audio_url you can listen to before committing.Get different options with --seed
The same prompt with the same seed always returns the same voices. Increment
--seed to explore new batches:Prompt tips
The quality of your voice depends on the quality of your description:| Prompt | Result |
|---|---|
"deep male voice with authority, like a movie trailer narrator" | Dramatic, resonant bass |
"friendly young woman, upbeat and energetic, American accent" | Casual, approachable |
"calm, measured British male, BBC documentary style" | Professional, trustworthy |
"enthusiastic tech reviewer, fast-paced, excited" | High energy, engaging |
"soft-spoken female, ASMR-like, soothing" | Gentle, intimate delivery |
Optional flags
| Flag | Description |
|---|---|
--gender | male or female — narrows results |
--locale | BCP-47 locale tag (e.g. en-US, pt-BR) for accent targeting |
--seed | Increment to get different batches (default: 0) |

