AI-901 - Implement AI solutions by using Microsoft Foundry - Section 2.2

Implement AI solutions for text and speech by using Foundry, including text analysis, spoken prompts to a multimodal model, and Azure Speech in Foundry Tools.

Building a lightweight application that includes text analysis; responding to spoken prompts by using a deployed multimodal model; and building a lightweight application using Azure Speech in Foundry Tools for speech-to-text and text-to-speech. The distinction to hold is between a multimodal model that accepts spoken input directly and the Azure Speech capability used for dedicated recognition and synthesis.

text analysis applicationmultimodal modelspoken promptsAzure Speech in Foundry Toolsspeech-to-texttext-to-speech

Practice question for this objective

Free sampleImplement AI solutions by using Microsoft Foundrymedium

A hospital wants a hands-free clinical note taker: as a doctor dictates observations into a microphone during a ward round, the application must convert the spoken words into written text that is saved to the patient record in real time. Which capability in Azure Speech in Foundry Tools meets this requirement?

  • AText-to-speech, which produces natural-sounding spoken audio from written input
  • BSpeech-to-text, which transcribes the spoken audio into written text as it is captured Correct
  • CSpeech translation, which converts spoken audio in one language into another language
  • DText analysis, which extracts sentiment and key phrases from a body of written text
Recognise that converting live spoken audio into written text is the speech-to-text capability of Azure Speech in Foundry Tools. Speech-to-text (speech recognition) takes an audio signal of spoken language and produces a written transcript, and it supports real-time streaming so words appear as they are spoken, which is what a live dictation workflow requires.

Why A is wrong: Text-to-speech runs in the wrong direction: it turns written text into audio, whereas the doctor is producing speech that must become text. It is tempting because both features live in Azure Speech, but it solves the reverse problem.

Why B is correct: The requirement is to turn dictated audio into written notes as they are spoken, which is exactly what speech-to-text does, supporting real-time transcription of a live microphone stream.

Why C is wrong: Speech translation is tempting because it also processes spoken input, but the scenario keeps a single language and only needs a transcript, not a change of language.

Why D is wrong: Text analysis operates on text that already exists and derives insights from it. It cannot capture spoken audio at all, so it fails the core dictation-to-text need despite sounding clinically relevant.

See more AI-901 practice questions, answers explained.

More in this domain

Back to all Implement AI solutions by using Microsoft Foundry objectives, or the AI-901 cert hub.

Examworthy is not affiliated with or endorsed by Microsoft. Original, blueprint-aligned practice material only.