AI-900 Practice Question: Describe features of Natural Language Processing workloads on Azure

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✗

  Extracting the most frequently used words from a speech transcript

  Why it's wrong here

  Word frequency analysis is a text analytics technique that operates on already transcribed speech, counting how often each word appears to reveal lexical patterns or topics. It is a batch or post-processing statistic, not a continuous audio-monitoring function, and it cannot activate or trigger any other processing. This is fundamentally different from keyword recognition, which is a lightweight, always-on acoustic detector that waits for a specific wake word.

• ✓

  Continuously listening for a specific wake word to activate full speech processing without cloud round-trips

  Why this is correct

  This describes the essential purpose of keyword recognition in Azure Speech: a tiny, always-on model on the device listens for a specific wake word, such as "Hey Azure," and only then initiates full speech-to-text processing in the cloud. This design eliminates continuous cloud round-trips, reducing bandwidth, latency, and cost while preserving privacy until the user explicitly invokes the assistant. It is exactly what the Azure Custom Keyword service supports, making this the correct explanation of the feature.

• ✗

  Highlighting important keywords in a speech transcript for meeting notes

  Why it's wrong here

  Highlighting important keywords in a meeting transcript is a natural language processing feature, typically implemented with key phrase extraction or named entity recognition on the full text to improve readability and summarization. It works after the speech has been transcribed, often in the cloud, and does not involve listening for a fixed acoustic trigger. Wake word recognition, by contrast, is a low-level microphone-front-end process that constantly checks for one pre-defined phrase, not a document annotation task.

• ✗

  Detecting when a customer uses specific product keywords during a support call

  Why it's wrong here

  Detecting product keywords during a support call is a speech analytics or conversation intelligence use case, where keyword spotting or phrase detection is applied to the call audio to assess compliance, sentiment, or topics. This analysis usually runs in the cloud on a recording or live stream and is not tied to device activation, whereas keyword recognition is a personal wake word mechanism optimized for on-device, power-efficient listening on a specific device. The scenario also implies detecting many possible words, whereas keyword recognition is trained for a single activation phrase.

Go deeper

Related to this question