AI-102 Question 222
Select 2You are building a voice-activated customer service application using Azure Speech Service. The application needs to recognize specific keywords such as 'help', 'cancel', and 'agent' to trigger appropriate actions. Which of the following steps are required to implement keyword recognition using Azure Speech SDK?
- A
Define a keyword recognition model using Azure Custom Speech and integrate it into the application.
- B
Use the Speech SDK to configure a keyword recognizer with predefined keyword models.
- C
Upload a list of keywords directly into the Azure portal for automatic recognition.
- D
Ensure the audio input is processed in real-time by the Azure Speech SDK for detecting keywords.
- E
Train a custom neural network using Azure Machine Learning to detect keywords.
Show answer and explanation
Correct answers: B, D
Explanation
To implement keyword recognition using Azure Speech Service, developers must utilize the Speech SDK to configure a keyword recognizer with predefined keyword models and ensure real-time audio processing. Azure provides built-in capabilities for keyword detection, eliminating the need for custom models or neural network training. Uploading keywords directly into the Azure portal is not a supported method in this context.
- A. Incorrect.
Defining a keyword recognition model using Azure Custom Speech is not required for keyword recognition. Azure Speech SDK provides predefined capabilities for keyword recognition, so custom models are not necessary.
- B. Correct.
This is correct. The Azure Speech SDK allows developers to configure a keyword recognizer with predefined keyword models to detect specific keywords.
- C. Incorrect.
This is incorrect. Keywords cannot be uploaded directly into the Azure portal for automatic recognition. Instead, the Speech SDK must be used to configure keyword detection.
- D. Correct.
This is correct. Real-time audio input processing is a critical component when using the Azure Speech SDK for keyword detection, as it ensures that keywords are detected as they are spoken.
- E. Incorrect.
This is incorrect. Training a custom neural network is not required for keyword recognition. Azure Speech SDK provides built-in support for keyword recognition without the need for custom neural network training.