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.

• ✓

  Create a Speech resource and obtain subscription key and region, then configure audio input, then initialize a SpeechRecognizer, then call the recognition method and process the result.

  Why this is correct

  This order is correct because the Azure Speech SDK requires a SpeechConfig object, built from the Speech resource's subscription key and region, before any recognition can be performed. Once those credentials exist, an AudioConfig (for microphone, file, or stream) is created, and both are passed into the SpeechRecognizer constructor. Only after the recognizer is fully initialized can you call a method such as RecognizeOnceAsync and process the resulting recognition text.

• ✗

  Configure audio input first, then create a Speech resource, then initialize a SpeechRecognizer, then call recognition.

  Why it's wrong here

  Configuring the audio input first is incorrect because the SpeechRecognizer cannot be constructed without a SpeechConfig that contains the subscription key and Azure region; the audio source is only one of the two required constructor inputs. Creating the Speech resource and obtaining its credentials is the foundational prerequisite, and doing audio configuration first does not initialize any valid service binding or make later steps possible.

• ✗

  Create a Speech resource, then initialize a SpeechRecognizer, then configure audio input, then call recognition.

  Why it's wrong here

  This ordering is invalid because a SpeechRecognizer binds its audio source at construction time; there is no supported way to attach or change AudioConfig after the recognizer object has already been created. Initializing SpeechRecognizer before configuring audio means the recognizer either uses a default microphone unintentionally or is missing required audio configuration, so recognition cannot proceed as intended.

• ✗

  Create a Speech resource, then call the recognition method, then initialize a SpeechRecognizer, then configure audio input.

  Why it's wrong here

  This sequence inverts the dependency order: the recognition method is an instance method of SpeechRecognizer, so it cannot be invoked before the recognizer object exists, and the recognizer cannot exist until both SpeechConfig (credential) and AudioConfig are provided. Calling the recognition method before initialization and audio configuration is logically impossible in the SDK and is a fundamental ordering error.

Go deeper

Related to this question