Skip to main content
In this quickstart, you create a voice-based prompt agent, open a live session, complete a spoken turn, and read the conversation back. Foundry Agent Service manages the voice orchestration, so you don’t host a speech pipeline of your own. Choose your development tool. The Foundry portal path creates and tests the agent in the browser with no code. The Python SDK and JavaScript/TypeScript SDK paths install the Microsoft Foundry SDK, create the agent in code, and connect to a live session over a WebSocket. The Azure Developer CLI path uses azd to scaffold, provision, deploy, and test the agent without writing runtime code. All paths create the same kind of managed prompt-based voice agent, which uses a managed real-time model.
Items marked (preview) in this article are currently in public preview. This preview is provided without a service-level agreement, and we don’t recommend it for production workloads. Certain features might not be supported or might have constrained capabilities. For more information, see Supplemental Terms of Use for Microsoft Azure Previews.
If you don’t have an Azure subscription, create a free account.

​
Prerequisites

The Foundry RBAC roles were recently renamed. Foundry User, Foundry Owner, Foundry Account Owner, and Foundry Project Manager were previously named Azure AI User, Azure AI Owner, Azure AI Account Owner, and Azure AI Project Manager. You might still see the previous names in some places while the rename rolls out. The role IDs and core permissions are unchanged by the rename.
  • A voice model available to your project, such as gpt-realtime. Confirm model availability on the Models page of your Foundry project.
  • Access to the prompt voice agent preview in your Azure subscription. Confirm that the preview is available in your project’s region before you start.

​
Related content