Azure Ai

Introduction

The Azure platform provides us with a pool of AI tools such as Azure OpenAI—a generative AI service—Azure AI Content Safety—for filtering harmful content—Azure AI Speech—a TTS or STT service—and much more. These AI services can be used in various environments, including web and mobile, without the hassle of implementing it from the ground up. Azure AI services follow three fundamental principles:

• Prebuilt and ready to use
• Accessed through APIs
• Available and secure on Azure

AI generally requires a lot of computing power and special knowledge. However, by making the service prebuilt, Microsoft eliminated the need for customers to actually have the hardware or expertise in AI, while leaving a room for customization.

Implementing the service on the API made it possible to easily access the service from client-side, and since the services are run securely on Azure, it promises consistent usability.

Azure AI service resources

There are two types of resources in Azure AI:

• Multi-service resource : a resource that provides access to multiple AI services with a single key and endpoint
• Single-service resources : a resource that provides access to a single AI service.

Once Azure AI service resources have been created, developers can choose from API, SDK, and VS interface.

Studio interfaces are very user friendly and provides an intuitive GUI approach to the beginner developers.

Authentication ensures that only permitted users have access to the AI services. Most of the time, users utilize RESTful API, which has a very concerete format and a mandatory key that inhibits from external being interrupting the transactions. Since the authentication also requires a specific endpoint, users must access their service in such a format : https://myaiservices29.cognitiveservices.azure.com/

Using Azure AI services

One can utilize Azure Machine Learning Studio to try out Azure’s newest AI interfaces.

While it is optimal to use API during usage, one can try out its features on the interface itself.

Endpoint and passkey are provided:

You can also set resources in order to inhibit connection:

And moderate content:

Reference

https://learn.microsoft.com/en-us/training/modules/fundamentals-azure-ai-services/