Skip to content

Latest commit

 

History

History
91 lines (69 loc) · 5.04 KB

readme.md

File metadata and controls

91 lines (69 loc) · 5.04 KB
Raw

Shop Easily Using Talking Avatar - Trailblazers Outdoor Gear

drawing

This solution accelerator can be used to deploy an application that offers an interactive shopping experience using a talking avatar. It uses Azure OpenAI combined with data stored on Cognitive Search and Azure SQL to generate answers.

Getting started

  1. Start by forking this repo

  2. Create the following Azure resources:

    • Azure OpenAI Service with these models deployed
      • gpt-35-turbo (note: version 0613 or higher is required)
      • text-embedding-ada-002 (version 2)
    • Azure AI Search with default settings
    • Azure SQL with the following settings
      • Authentication: SQL and Microsoft Entra authentication enabled
      • Networking: Allow Azure services and resources to access this server enabled
    • Azure Speech Service
    • Azure AI services multi-service account
    • Azure Blob Storage account

  3. Upload the images in the product-images directory to a blob container in the Storage Account. Generate a SAS URL for the blob storage container. Set the expiry date according to the planned lifecycle of your application.

    drawing

  4. Run the cells in the create-index-and-database.ipynb notebook to upload the product data to Azure AI Search and the Azure SQL Database.

  5. Create a file named local.settings.json in the api directory of the repository. Make sure to add the following variables to local.settings.json. The AzureWebJobsStorage variable can be left empty for development purposes.

{
  "IsEncrypted": false,
  "Values": {
    "AzureWebJobsStorage": "",
    "FUNCTIONS_WORKER_RUNTIME": "python",
    "AZURE_OPENAI_ENDPOINT": "https://XXX.openai.azure.com/",
    "AZURE_OPENAI_API_KEY": "XXX",
    "AZURE_OPENAI_CHAT_DEPLOYMENT" : "gpt-35-turbo-16k",
    "AZURE_OPENAI_EMBEDDINGS_DEPLOYMENT" : "text-embedding-ada-002",
    "AZURE_OPENAI_API_VERSION" : "2023-07-01-preview",
    "AZURE_SEARCH_ENDPOINT": "https://XXX.search.windows.net",
    "AZURE_SEARCH_API_KEY": "XXX",
    "AZURE_SEARCH_INDEX": "products",
    "AZURE_SPEECH_REGION": "westeurope",
    "AZURE_SPEECH_API_KEY": "XXX",
    "TEXT_ANALYTICS_ENDPOINT": "XXX",
    "TEXT_ANALYTICS_KEY": "XXX",
    "BLOB_SAS_URL": "https://XXX",
    "SQL_DB_SERVER": "XXX.database.windows.net",
    "SQL_DB_USER": "XXX",
    "SQL_DB_PASSWORD": "XXX",
    "SQL_DB_NAME": "OutdoorEquipmentShop"
  }
}

  1. In case you are using an Azure Speech Services instance in a region different from westeurope, update line 17 of main.js in the src/js folder to reflect that.

  2. This application can be deployed using Azure Static Web Apps. Refer to this quickstart to learn more. This application is using no front-end frameworks.

    If you are using Visual Studio Code, you can execute the following steps:

    • Install the Azure Static Web Apps and Azure Functions extensions

    • Right-click on Static Web Apps extension folder, select Create Static Web App ... (Advanced) with the following parameters:

      Parameter Description
      Resource group Select an existing resource group or create a new one
      Name Choose a name, e.g., avatar-app
      Pricing option Standard
      Region Select the same or a nearby region as for the above resources
      Framework Custom
      Application code location avatar/interactive/src
      Build output location (Leave blank)

  3. In the VSCode Static Web Apps extension, navigate to Application Settings of your app and right-click Upload Local Settings. This will populate the settings from local.settings.json to the web app.

  4. In the VSCode Static Web Apps extension, right-click on your app name and select Browse site to use the app

Notes on running the solution locally

  • ODBC Driver 17 for SQL Server is required to run the solution locally.
  • Use the Static Web Apps CLI to run the solution. After navigating in the terminal to avatar/interactive, the following command can be used to run the solution: swa start src --api-location api.
  • The solution has been tested with Node version 18.0.0.

Hints on debugging

  • The login screen is currently non-functional. If you click on 'login' without entering any information, you will be redirected to the main page.
  • If the avatar is not loading on the main page, refresh the web page with the console open. This will show the error message.