Release2026-09-07

Gradium launched Voice Design, which creates new synthetic voices from a written description. Users write one or two sentences and get several candidate voices back in seconds. A kept voice runs on the same text-to-speech endpoint as any catalog voice, with the same latency. The feature is live in Gradium's API and Studio, and free to use. Supported languages are English, French, German, Spanish and Portuguese, with regional accents. Gradium says designed voices are fully synthetic, so no voice actor licence or royalty applies. It ran blind listening tests against ElevenLabs, Inworld, MiniMax and Fish Audio, and reports first place in every language. A Gemini 3.1 Pro judge produced the same ranking. The comparison figures come from Gradium's own testing.

What changed

New voices required sourcing, auditioning and licensing a real speaker to clone.

What it unlocks

Creating a permanent synthetic voice from a written description in seconds, with no speaker licence.

  • 72.6% win rate on accent prompts
  • 13.6 points clear of next system
  • 7,627 blind comparisons
  • up to 5 candidate voices per prompt

What you need to act on it

  • a Gradium account, via the API or Studio

Send this to someone who needs it

Shares the story and its sources. Nothing about you.

What does this mean for your job?

This is the story as everyone gets it. Once a week we send you the version written for your role — what changed, why it matters for the work you actually do, and one thing to try. Free while we tune it.