What does speech-to-speech translation in Azure Speech do?

Converts text in one language to text in another language

Converts text to speech in the same language

Converts spoken language in one language into spoken language in another language

Converts speech to text in the same language

A navigation app reads street names aloud. Users report that some street names are mispronounced. You need to fix the pronunciation without changing the voice. What should you do?

Use the Speech SDK's translation to translate street names to a different language.

Use the Speech SDK's keyword recognition to detect street names and play pre-recorded audio.

Use SSML with elements for the problematic street names.

Use a different neural voice that is known for better street-name pronunciation.

A university is building a lecture transcription service. Professors have strong accents and use domain-specific terms. The service must transcribe live lectures with low latency. You have a small set of labeled audio samples from each professor. What should you do to maximize accuracy?

Train a custom speech model with audio from all professors and use it in real-time recognition.

Use the standard speech model with language models for each professor.

Use the Speech SDK's translation feature to translate lectures to a standard dialect.

Use batch transcription with a custom model and post-process with a language model.

A home automation system uses Azure Speech. It must wake up on 'Hey Home' and then understand commands like 'Turn on the lights'. The system should only process commands after the wake word. What should the developer implement?

Use Speech-to-Text continuously and filter for 'Hey Home'.

Use Custom Speech to train a model that outputs 'Hey Home'.

Use KeywordRecognizer for 'Hey Home' and then Intent Recognition for the command.

Use Intent Recognition with a pattern for 'Hey Home'.

A hospital uses Azure Speech to transcribe doctor-patient conversations. The transcription must be accurate for medical terms like 'myocardial infarction'. The team has a set of recorded medical conversations with transcripts. What should they do to improve accuracy?

Use Custom Speech to train a model with the medical audio and transcripts.

Use SSML to add pronunciation hints for medical terms.

Use Speech Translation to translate the conversation to a standard language.

Use Intent Recognition to classify the medical terms.

Microsoft Certified:Azure AI Engineer Associate

Domain 5Objective 2

Process and Translate Speech AI-102 Practice Questions (Page 9)

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 41–45

41foundation · easy
What does speech-to-speech translation in Azure Speech do?
Select an answer first
42application · medium
A navigation app reads street names aloud. Users report that some street names are mispronounced. You need to fix the pronunciation without changing the voice. What should you do?
Select an answer first
43expert · hard
A university is building a lecture transcription service. Professors have strong accents and use domain-specific terms. The service must transcribe live lectures with low latency. You have a small set of labeled audio samples from each professor. What should you do to maximize accuracy?
Select an answer first
44application · medium
A home automation system uses Azure Speech. It must wake up on 'Hey Home' and then understand commands like 'Turn on the lights'. The system should only process commands after the wake word. What should the developer implement?
Select an answer first
45application · medium
A hospital uses Azure Speech to transcribe doctor-patient conversations. The transcription must be accurate for medical terms like 'myocardial infarction'. The team has a set of recorded medical conversations with transcripts. What should they do to improve accuracy?
Select an answer first

Finished these 5 questions?

Review the revealed explanations, or continue through the curriculum.

No more pagesBack 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.