Azure AI Fundamentals · 57% of the exam

Implement AI solutions by using Microsoft Foundry: free practice questions

5 sample questions from our 55-question bank for this domain — answers and explanations included. These are the same scenario-based style as the real Azure exam.

1. You are building a voice-response feature with Azure Speech and want the synthesized voice to sound natural rather than robotic, matching your brand. What should you configure?

  • A. Select an appropriate neural voice (and language/locale) for the text-to-speech output✓ Correct
  • B. Increase the audio file's resolution in pixels
  • C. Switch entirely to a computer vision service instead
  • D. Disable text-to-speech and only ever return raw text
Explanation

Azure Speech's text-to-speech offers neural voices that sound natural and can be selected/configured for language and style to match a desired brand voice. Audio has no 'pixel resolution', switching to computer vision is irrelevant to audio output, and disabling text-to-speech removes the very feature being requested.

2. While testing a single-agent solution in the Foundry portal, you want to see exactly which tool the agent called and what arguments it passed for a given user turn. Where would you look?

  • A. The run/thread trace or step details shown for that conversation turn in the portal✓ Correct
  • B. The Azure Cost Management dashboard
  • C. The regional availability list for the model
  • D. The subscription's IAM role assignments page
Explanation

The portal exposes the run or thread details for an agent conversation, including which tool calls were made and with what arguments, so you can debug behavior. Cost Management, regional availability, and IAM role pages are unrelated administrative views that do not show per-turn tool call details.

3. In the Foundry portal, what is the purpose of the 'instructions' field when configuring a new single-agent solution?

  • A. It defines the agent's persistent role, behavior, and constraints that guide it across the whole conversation✓ Correct
  • B. It stores the transcript of the most recent conversation only
  • C. It sets the Azure subscription billing tier for the agent
  • D. It lists the names of other unrelated agents in the project
Explanation

An agent's instructions function like a system prompt: they define its role and behavioral constraints for the entire session, not just one exchange. They are not a transcript log, they do not control billing tier, and they are not a directory of other agents.

4. When extracting information from a video that includes both spoken dialogue and on-screen text (like a lower-third caption), what should you expect from Azure Content Understanding?

  • A. It can potentially extract both the spoken audio transcript and the visible on-screen text, depending on how the analyzer is configured✓ Correct
  • B. It can only ever process the video's audio track and completely ignores all visual frames
  • C. It can only ever process the visual frames and completely ignores all audio
  • D. Video content is entirely unsupported by Content Understanding
Explanation

Content Understanding is designed to work across modalities present in video — audio for spoken dialogue and visual frames for on-screen text — with the analyzer configuration determining which are extracted. Claiming it only handles audio, only handles visuals, or does not support video at all understates its actual multimodal extraction capability.

5. A marketing team wants to generate several new product-concept images from a text description using a deployed generative model in Foundry. Which capability are they using?

  • A. Image generation from a generative model, producing new visual outputs from a text prompt✓ Correct
  • B. Text analysis to summarize existing product reviews
  • C. Speech-to-text transcription of a marketing meeting
  • D. Content Understanding form extraction from a scanned invoice
Explanation

Creating new images from a text description is image generation using a generative model — exactly the described use case. Text analysis, speech-to-text, and document form extraction are all different capabilities that operate on existing text, audio, or documents rather than creating new images.

50 more questions in this domain

Practice the full bank with instant grading, flashcards, and a timed mock exam.

Start practicing free