Azure Speech in Foundry Tools

  • 2 minutes

Azure Speech in Foundry Tools is a set of speech-related capabilities that are provided by a Foundry resource. You can use these capabilities to add speech support to apps and agents built in Microsoft Foundry projects. For example:

  • Creating an application to transcribe recorded calls or meetings.
  • Creating an AI assistant that can read text messages or emails aloud.

Diagram showing an Azure Speech resource performing speech-to-text and text-to-speech functions.

Using Azure Speech in a Microsoft Foundry resource

To use Azure Speech in Foundry Tools, you must provision a Microsoft Foundry resource in your Azure subscription.

After you have provisioned a Foundry resource in your Azure subscription, you can use its endpoint to call the Azure Language APIs from your code, authenticating requests by providing the key associated with your resource. You can call the Azure Language APIs by submitting requests in JSON format to the REST interface, or by using any of the available programming language-specific SDKs.

Note

The code examples in this module are based in Python, using the Python SDK for Azure Speech in Foundry Tools. SDKs for other common languages (such as Microsoft C#, JavaScript, and others) follow a similar pattern.

Creating a SpeechConfig

The initial object you need to create to provide access to the Azure Speech in Foundry Toole endpoint is a SpeechConfig object; which encapsulates the connection details for the service in your Foundry resource.

Tip

The default home page in the Foundry portal shows the endpoint and key for your project. To view the key and endpoint for your resource, you can view the parent resource for your project in the Admin tab of the Operate page of the portal. The project and foundry resource keys are the same, and the project endpoint is the resource endpoint with /api/projects/{project_name} appended - so if the project endpoint is https://my-ai-app-foundry.services.ai.azure.com/api/projects/my-ai-app, then the resource endpoint is https://my-ai-app-foundry.services.ai.azure.com.

For example, the following Python code creates a SpeechConfig object that can be used to submit requests to Azure Speech APIs in a Foundry resource.

# run "pip install azure-cognitiveservices-speech" first to install the package 
import azure.cognitiveservices.speech as speech_sdk

# Create SpeechConfig using endpoint and key
speech_config = speech_sdk.SpeechConfig(subscription="YOUR_FOUNDRY_KEY",
                                        endpoint="YOUR_FOUNDRY_ENDPOINT")

Note

Releases of the Python SDK prior to 1.48.2 required that you specify the region where your resource is deployed instead of the endpoint. With the latest release, you can use either the Foundry resource endpoint or the region.