version: "1.0.0" name: design-voice description: Create a brand-new voice from a text description, preview candidates, and publish one as a reusable voice. license: MIT

Skill: design-voice

Design a voice that doesn't exist yet from a plain-text description ("a warm, gravelly late-night radio host with a slow cadence"), preview candidates, then publish the one you like as a reusable voice for TTS. Voice Design is part of the AudioSonic voice engine.

When to use

Invoke when the user wants:

• An original character or brand voice, without a reference recording
• Several voice options to choose from before committing
• A reusable custom voice they can call from text-to-speech

1. Generate previews

POST https://api.audiopod.ai/api/v1/voice/design

Headers: X-API-Key: ap_your_api_key

Body:

{
  "design_prompt": "a warm, gravelly late-night radio host, slow and intimate",
  "preview_text": "You're listening to the midnight hour.",
  "num_samples": 3
}

design_prompt is 30–250 chars, preview_text is 20–300 chars, and num_samples is 1–3. Returns candidate previews. This is a paid-tier feature, billed per preview delivered (and refunded on failure).

2. Publish a preview

POST https://api.audiopod.ai/api/v1/voice/design/publish

Publishes a chosen preview into a durable voice_id you can reuse with the text-to-speech skill.

Errors

• 402 PREMIUM_TIER_REQUIRED — voice design requires a paid tier

Pricing

Billed per preview (250 credits/preview). See https://www.audiopod.ai/developers/pricing.