ElevenLabs
How the ElevenLabs integration powers spoken voiceovers and Voice Agent conversations inside Stated publications.
· 1 min read
Stated uses ElevenLabs to power spoken voice features in Pages and Experiences. You do not need to build a microphone interface, transcript store or results dashboard yourself: Stated handles the publication flow around the conversation.
What ElevenLabs powers
ElevenLabs provides speech synthesis and the voice conversation layer used by Stated Voice Agents. A Voice Agent can greet a visitor, ask configured questions, respond in the selected language and guide the conversation toward its purpose.
Stated stores the resulting conversation as a response on the publication. It does not turn the raw provider session into a separate public asset.
Create a Voice Agent
Add a Voice Agent to a Page or Experience, then choose:
- a name and visible call to action;
- an opening message;
- a purpose: answer questions, interview the visitor or collect feedback;
- instructions and ordered questions;
- voice, language and maximum duration.
Write instructions for the visitor's job, not for the model's internal process. For example: “Ask these three customer-research questions one at a time. Do not claim facts that are not in the publication.”
Privacy and results
A visitor is asked for microphone permission only when they start. When identity collection is enabled, Stated asks for name and email before the conversation. Completed transcripts and answers are visible to the publication owner in Results, alongside other interactions.
Plan limits
Voice usage consumes the account's available voice minutes. Stated checks entitlement before the conversation and applies visitor, publication and account limits. If the account has no remaining allowance, the voice call does not start.
For the complete visitor and results behaviour, see Voice Agents.