Exercise - Configure Azure OpenAI Service to generate information about your podcast

  • 10 minutes

Background

In this exercise, you explore Azure OpenAI Service. You learn about the service and models, and then you go through the process of creating deployments. This exercise establishes the basis for seamless AI integration.

Definition: Azure OpenAI Service

Azure OpenAI Service provides REST API access to the powerful language models of OpenAI, including the GPT-4, GPT-4 Turbo with Vision, GPT-3.5-Turbo, and Embeddings model series. Additionally, the new GPT-4 and GPT-3.5-Turbo model series have now reached general availability. You can adapt these models to your specific tasks, including content generation, summarization, image understanding, semantic search, and natural language-to-code translation. Users can access the service through REST APIs, Python SDK, or the web-based interface in the Azure OpenAI Studio.

For more information, see What is Azure OpenAI Service?

Prerequisites

For this exercise, make sure that you meet the following prerequisites:

Exercise steps

The following video goes through the steps for this unit's exercise.

Create the Azure OpenAI Service resources

To create the Azure OpenAI Service resources, follow these steps:

  1. Go to portal.azure.com and sign in with your Azure account.

  2. Search for and select Resource groups.

  3. Select Create.

    Screenshot of the Create button to create a resource group.

  4. Name your resource group PodcastCopilotPowerApp, select Review + Create, and then select Create.

  5. Create a new Azure OpenAI Service resource and then configure the following parameters:

    • Name - podcastcopilotwe-{your initials}, for example podcastcopilotwe-jd

    • Subscription - Select your Azure subscription

    • Resource group - Select the resource group that you created in the previous step

    • Location - West Europe

    • Pricing tier - Select the Standard S0 pricing tier

    You use this resource to access the Whisper and GPT-3.5 models.

    Screenshot of the Create Azure Open A I Service resource page with parameters filled in.

  6. Select Next until you get to the Review + Submit page, and then select Create to create the resource.

  7. Create another new Azure OpenAI Service resource, and then configure the following parameters:

    • Name - podcastcopilotsc-{your initials}, for example podcastcopilotsc-jd

    • Subscription - Select your Azure subscription

    • Resource group - Select the resource group that you created in the previous step

    • Location - Sweden Central

    • Pricing tier - Select the Standard S0 pricing tier

    You use this resource to access the Dall.E 3 model.

  8. Select Next until you get to the Review + Submit page, and then select Create to create the resource.

Create the Bing Search resource in Azure

To create the Bing Search resource in Azure, follow these steps:

  1. Create a Bing Search resource in Azure, and then configure the following parameters:

    • Name - podcastcopilotbing-{your initials}, for example podcastcopilotbing-jd

    • Subscription - Select your Azure subscription

    • Resource group - Select the resource group that you created in the previous step

    • Region - Global

    • Pricing tier - Select the F1 pricing tier

  2. Read the Terms and Notice and then select the checkbox to agree to them.

    Screenshot of the terms confirmation for Bing Search.

  3. Select Review + Create, and then select Create to create the resource.

  4. Open your PodcastCopilotPowerApp resource group in the Azure portal.

  5. Select the podcastcopilotbing-{your initials} resource.

  6. Select Keys and Endpoint on the left menu and then note the Key 1 value because you need it for subsequent steps.

    Screenshot of the Bing Search key.

Create model deployments in Azure OpenAI Studio

To create model deployments in Azure OpenAI Studio, follow these steps:

  1. Open your PodcastCopilotPowerApp resource group in the Azure portal.

  2. Select the podcastcopilotwe-{your initials} resource.

  3. From the upper menu, select Go to Azure OpenAI Studio.

    Screenshot of the Go to Azure OpenAI Studio button.

  4. In Azure OpenAI Studio, select Deployments on the left menu.

  5. Select + Create New Deployment, and then configure the following parameters:

    • Select a model - GPT-35-Turbo

    • Model version - Auto-update to default

    • Deployment name - gpt35turbo

  6. Select Create to create the deployment.

    Screenshot of the Deploy model dialog with details filled in.

  7. Select + Create New Deployment again, and then configure the following parameters:

    • Select a model - Whisper

    • Model version - Auto-update to default

    • Deployment name - whisper

  8. Select Create to create the deployment.

  9. Select the gpt35turbo deployment and then select Open in Playground.

    Screenshot of the Open in Playground button.

  10. In the playground, select View Code. Note the following information because you need it for later steps:

    • Deployment Name - gpt35turbo

    • Endpoint - https://podcastcopilotwe-{your initials}.openai.azure.com/

    • Key - xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

    Screenshot of the copy information buttons in the Sample Code dialog.

  11. When you complete the process, repeat it for the whisper deployment.

  12. Select Deployments on the left menu, select the whisper deployment. Then, select Open in Playground.

  13. Scroll down to the Next Steps section, select the podcastcopilotwe-{your initials} resource, and then note the following information because you need it for subsequent steps:

    • Deployment Name - whisper

    • Endpoint - https://podcastcopilotwe-{your initials}.openai.azure.com/

    • Resource Key - xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

  14. In the upper-right corner, select your name. Then, under Current resource, select Switch.

    Screenshot of the Switch button under Current resource.

  15. Select the podcastcopilotsc-{your initials} resource and then select Use resource.

    Screenshot of the Use resource button.

  16. Select Deployments on the left menu, select Create New Deployment, and then configure the following parameters:

    • Select a model - dall-e-3

    • Model version - Auto-update to default

    • Deployment name - dalle3

  17. Select Create to create the deployment.

  18. Select the dalle3 deployment and then select Open in Playground.

  19. Select View Code and then note the following information because you need it for later steps:

    • Deployment Name - dalle3

    • Endpoint - https://podcastcopilotsc-{your initials}.openai.azure.com/

    • Key - xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

Upload a podcast snippet to Azure Blob Storage

In this exercise, you use Microsoft Azure Blob Storage to store the podcast snippet audio file. This step allows you to easily test your API and application with a real-world audio file.

  1. Go to the PodcastSnippet.mp3 file in this repository and download it to your local machine.

    Screenshot of the download podcast snippet button.

  2. Go to portal.azure.com and sign in with your Azure account.

  3. Go to Create a storage account and set the following parameters:

    • Subscription - Select your Azure subscription

    • Resource group - Select PodcastCopilotPowerApp

    • Storage account name - Enter podcaststorage{your initials}, such as podcaststoragejd, to make it unique

    • Location - East US

    • Performance - Standard

    • Redundancy - Geo-redundant storage (GRS)

    • Ensure that Make read access to data available in the event of regional unavailability is selected

  4. Select Review + Create and then select Create to create the storage account.

  5. After you create the storage account, select Go to resource.

    Screenshot of the Go to resource button.

  6. On the left panel, scroll to the Settings section and select Configuration.

  7. Search for the Allow Blob anonymous access option and set it to Enabled.

    Screenshot of the Allow Blob anonymous access option set to Enabled.

  8. Select Save in the upper part of the page to save the changes.

  9. After the changes save, scroll to the Data Storage section, select Containers > + Container. Then, enter the following information.

    • Name - audio

    • Anonymous access level - Blob (anonymous read access for blobs only)

  10. Select Create to create the container.

  11. After you create the container, select the audio container and then select Upload.

  12. Upload the PodcastSnippet.mp3 file that you downloaded earlier and then select Upload.

    Screenshot of the Upload button.

  13. When the file uploads, select the PodcastSnippet.mp3 file and then copy the URL value. Note this value because you need it for later steps.

Next steps

Now, you've learned how to generate information about a recorded podcast by using Azure OpenAI Service. Next, you learn how to bridge the gap to Microsoft Power Platform with custom connectors and .NET API.