ElevenLabs offers Voice Design, a tool launched in February 2023 that generates entirely new AI voices from a text description, allowing users to create a voice that has no real-world equivalent without uploading any audio recordings. The tool is available to all registered users, including those on the free tier, and produces three unique voice options per prompt for preview and selection.

Voice Design takes a natural language prompt describing desired vocal characteristics, such as gender, age, accent, and personality, and generates three unique synthetic voice options as output. Each generation introduces randomness so that voices produced from the same prompt are distinct from one another. The tool is positioned as a supplement to the Voice Library for cases where no existing voice fits a project. Generated voices can be previewed and saved to a user's voice library for use in subsequent text-to-speech generation. The tool is also accessible via the ElevenLabs API.