A wearable device uses Azure Speech to detect a specific emergency phrase like 'Help me'. The device has limited battery and processing power. The phrase must be detected quickly and reliably, even in noisy environments. What is the best approach?

Use the KeywordRecognizer with a custom keyword model trained on noisy audio.

Use continuous Speech-to-Text and filter for the phrase.

Use Intent Recognition to detect the phrase.

Use Speech Translation to translate the phrase into a standard command.

A developer is using Azure Text-to-Speech to generate audio for an e-learning platform. They want to emphasize certain words and add a pause between sections. What should they use?

Use a different neural voice that naturally emphasizes words.

Add exclamation marks and commas to the text.

Use Custom Speech to train the model to emphasize words.

A healthcare app reads aloud medication instructions to elderly patients. The app currently uses a neural voice with default SSML. Users complain that the speech sounds robotic and that the app sometimes mispronounces the drug name 'metoprolol'. You need to improve naturalness and pronunciation without changing the voice. What should you do?

Add a element after each syllable of the drug name and use on the first syllable.

Use the element with style='chat' and increase the speaking rate.

Add a element with the correct IPA for 'metoprolol' and use to vary pitch and rate.

Switch to a different neural voice that is preconfigured for medical terminology.

A global company runs multilingual meetings. They want a real-time application that listens to a speaker in English and outputs spoken Spanish to participants. The application must also display a Spanish text transcript for accessibility. What should the developer use?

Custom Speech with a Spanish language model.

Speech-to-Text with Spanish translation, then feed the text to Text-to-Speech.

Speech-to-Text in English and then use a separate translation API.

Speech Translation with speech output set to Spanish.

Microsoft Certified:Azure AI Engineer Associate

Domain 5Objective 2

Process and Translate Speech AI-102 Practice Questions (Page 6)

Part of the Implement natural language processing solutions domain, which accounts for 15-20% of the AI-102 exam. Microsoft does not publish an official question count, but from its 100-minute exam (~40–65 total, ~6–13 in this domain), expect 2–4 from this objective — we provide 45 practice questions to prepare you well beyond it. (estimate)

45questions here

9free pages

9concepts

15-20%of the exam

Questions 26–30

26expert · hard
A wearable device uses Azure Speech to detect a specific emergency phrase like 'Help me'. The device has limited battery and processing power. The phrase must be detected quickly and reliably, even in noisy environments. What is the best approach?
Select an answer first
27application · medium
A developer is using Azure Text-to-Speech to generate audio for an e-learning platform. They want to emphasize certain words and add a pause between sections. What should they use?
Select an answer first
28application · medium
A healthcare app reads aloud medication instructions to elderly patients. The app currently uses a neural voice with default SSML. Users complain that the speech sounds robotic and that the app sometimes mispronounces the drug name 'metoprolol'. You need to improve naturalness and pronunciation without changing the voice. What should you do?
Select an answer first
29foundation · easy
Which Azure Speech feature enables real-time translation of a spoken conversation into another spoken language?
Select an answer first
30application · medium
A global company runs multilingual meetings. They want a real-time application that listens to a speaker in English and outputs spoken Spanish to participants. The application must also display a Spanish text transcript for accessibility. What should the developer use?
Select an answer first

Finished these 5 questions?

Review the revealed explanations, or continue through the curriculum.

Next 5 questions →Back to AI-102

Free Basic Practice is a study aid with revealable answers — not a scored exam. Examers.io is independent and not affiliated with or endorsed by Microsoft. “AI-102” is a trademark of its owner, used for identification only.

Process and Translate Speech AI-102 Practice Questions (Page 6)

Questions 26–30

A wearable device uses Azure Speech to detect a specific emergency phrase like 'Help me'. The device has limited battery and processing power. The phrase must be detected quickly and reliably, even in noisy environments. What is the best approach?

A developer is using Azure Text-to-Speech to generate audio for an e-learning platform. They want to emphasize certain words and add a pause between sections. What should they use?

Which Azure Speech feature enables real-time translation of a spoken conversation into another spoken language?

A global company runs multilingual meetings. They want a real-time application that listens to a speaker in English and outputs spoken Spanish to participants. The application must also display a Spanish text transcript for accessibility. What should the developer use?

Keep studying — Implement natural language processing solutions