Share via

Test: Azure Speech Service Not Available in Foundry Tools

Abhinava Maddha 105 Reputation points Microsoft Employee
2026-04-28T21:01:13.8433333+00:00

Question: I’m attempting to use Azure Speech services through Foundry tools, but the option does not show up or fails during setup.

Details:

  • Azure Speech resource is active
  • Using Foundry Tools interface
  • No clear error message, just missing or unavailable

Questions:

  1. Why might Azure Speech not appear as an available tool in Foundry?
  2. Does it require specific regions or configurations to work with Foundry?
  3. How can I properly connect or enable Speech services within Foundry?
Azure Speech in Foundry Tools
0 comments No comments

Answer accepted by question author

  1. Q&A Assist
    2026-04-28T21:01:25.3866667+00:00
    AI generated content. This question contains an answer created with AI from Q&A Assist. Learn more
    1. Azure Speech might not appear as an available tool in Foundry when:
      • The Foundry resource is created in a region that does not support Azure Speech in Foundry Tools or MCP tools. In that case, the Azure Speech MCP Server will not show in the tool catalog.
      • The scenario or tenant does not support the Speech MCP tool. In this case, the tool is simply not listed in the catalog.
    2. Azure Speech does require specific regions and configurations to work with Foundry:
      • A Microsoft Foundry resource must be created in a supported region for Speech. Speech capabilities in Foundry Tools follow the Speech service supported regions.
      • The Speech MCP Server is only available in regions where Foundry Agent Service supports MCP tools. Regional availability for Foundry features, including Speech capabilities and Agent Service, is documented under Foundry feature availability.
      • The Speech MCP tool does not support Network-secured Microsoft Foundry. If the Foundry resource is network-secured, the Speech tool will not be usable.
    3. To properly connect or enable Speech services within Foundry:
      1. Verify prerequisites:
        • Ensure there is an active Azure subscription.
        • Create a Microsoft Foundry resource in a supported Speech region.
        • Ensure an Azure Storage account exists with at least one blob container for audio input/output, and that the account has the Storage Blob Data Contributor role so SAS URLs can be created.
      2. Create or open an agent in Foundry:
        • Go to Microsoft Foundry → BuildAgentsCreate agent (or open an existing agent).
      3. Add the Azure Speech tool:
        • Open the agent playground.
        • Under Tools, select Add → Add a new tool.
        • In Select a tool, choose the Catalog tab.
        • Search for Azure Speech MCP Server, select it, then select Create.
        • If the tool does not appear here, confirm the Foundry resource is in a supported region for Speech and MCP tools.
      4. Configure the tool:
        • Parameters → foundry-resource-name: Enter the name of the Foundry resource created in the prerequisites.
        • Authorization → Bearer (API Key): Use KEY1 or KEY2 from the Foundry resource Keys and Endpoint page.
        • Authorization → X-Blob-Container-Url: Generate a container SAS URL with read and write permissions and a valid expiry time, and paste it here.
      5. Connect and validate:
        • Select Connect to add the remote Speech MCP server as a tool.
        • After a successful connection, the Speech tool appears in the agent’s Tools list with a connected status.

    If connection fails or audio links do not work:

    • Recopy the API key (KEY1 or KEY2) from the Foundry resource if authorization errors occur.
    • Regenerate the container SAS URL with correct permissions and expiry if speech output audio links fail.
    • Regenerate the file SAS URL if speech-to-text cannot access the audio file.

    References:

    1 person found this answer helpful.

0 additional answers

Sort by: Most helpful

Your answer

Answers can be marked as 'Accepted' by the question author and 'Recommended' by moderators, which helps users know the answer solved the author's problem.