Explanation:

In machine learning, especially within the Microsoft Azure AI Fundamentals (AI-900) framework, datasets used for supervised learning are composed of features (inputs) and labels (outputs). According to the Microsoft Learn module "Explore the machine learning process", a feature is any measurable property or attribute used by the model to make predictions, whereas a label is the actual value or category the model is trying to predict.
* Household Income # FeatureA feature (also known as an independent variable) represents the input data that the machine learning algorithm uses to detect patterns or correlations. In this dataset, Household Income is a numeric value that influences the prediction of house price categories. During training, the model learns how variations in household income correlate with changes in the house price category.
Microsoft Learn defines features as "the attributes or measurable inputs that are used to train the model." Thus, Household Income serves as a predictive input or feature.
* House Price Category # LabelThe label (or dependent variable) represents the output the model aims to predict. It is the known result during training that helps the algorithm learn correct mappings between features and outcomes. In this scenario, House Price Category-which can take values such as "Low,"
"Middle," or "High"-is the classification outcome that the model will predict based on household income (and possibly other variables). According to Microsoft Learn, "the label is the variable that contains the known values that the model is trained to predict." In summary, the dataset defines a supervised learning classification problem, where Household Income is the feature (input) and House Price Category is the label (output) that the model will learn to predict.

Explanation:

"Generative AI enables software applications to generate new content, such as language dialogs and images."
- YES
This statement is true. According to the Microsoft Azure AI Fundamentals (AI-900) study guide and Microsoft Learn documentation, Generative AI refers to systems capable of creating new content such as text, audio, images, video, and code. Models like GPT, DALL E, and Codex use deep learning to generate human- like responses, natural conversations, or creative media. This is a key differentiator between generative and discriminative AI - generative AI produces new data, while discriminative AI categorizes or analyzes existing data.
"The difference between a large language model (LLM) and a small language model (SLM) is the number of variables in the model." - YES This statement is true. The primary distinction between an LLM and an SLM lies in the scale of parameters (variables) within the neural network. LLMs contain billions or even trillions of parameters, which enable them to capture complex linguistic patterns and perform broader tasks. SLMs have fewer parameters, making them faster but less capable of handling complex, context-rich tasks.
"Generative AI is a type of supervised learning." - NO
This statement is false. Generative AI models are typically trained using unsupervised or self-supervised learning methods. They learn by predicting missing or next elements in large text or image datasets rather than relying on labeled input-output pairs, which are used in supervised learning.

According to the Microsoft Azure AI Fundamentals (AI-900) official study guide and the Microsoft Learn module "Identify features of speech capabilities in Azure Cognitive Services", the Azure Speech service provides functionality for converting text to spoken words (speech synthesis) and speech to text (speech recognition).
In this scenario, the app must read recipe instructions aloud to assist users with visual impairments. This task is achieved through speech synthesis, also known as text-to-speech (TTS). The Azure Speech service uses advanced neural network models to generate natural-sounding voices in many languages and accents, making it ideal for accessibility scenarios such as screen readers, virtual assistants, and educational tools.
Microsoft Learn defines Speech service as a unified offering that includes:
* Speech-to-text (speech recognition): Converts spoken words into text.
* Text-to-speech (speech synthesis): Converts written text into natural-sounding audio output.
* Speech translation: Translates spoken language into another language in real time.
* Speaker recognition: Identifies or verifies a person based on their voice.
The other options do not fit the requirements:
* A. Text Analytics - Performs text-based natural language analysis such as sentiment, key phrase extraction, and entity recognition, but it cannot produce audio output.
* B. Translator Text - Translates text between languages but does not generate speech output.
* D. Language Understanding (LUIS) - Interprets user intent from text or speech for conversational bots but does not read text aloud.
Therefore, based on the AI-900 curriculum and Microsoft Learn documentation, the correct service for converting recipe text to spoken audio is the Azure Speech service.
# Final answer: C. Speech
Reference:Microsoft Learn - "What is the Speech service?" (Azure Cognitive Services > Speech)

Explanation:

According to the Microsoft Azure AI Fundamentals (AI-900) official study guide and the Microsoft Learn module "Identify features and uses of speech capabilities", speech recognition refers to the process of converting spoken words into written text. When a speaker's voice is transcribed into subtitles during a presentation, the system listens to the audio input, identifies the spoken words, and generates corresponding text in real time. This is precisely what speech recognition technology accomplishes.
Azure provides this functionality through the Azure Speech Service, which supports multiple speech-related features:
* Speech-to-Text (Speech Recognition) - Converts spoken audio into text.
* Text-to-Speech (Speech Synthesis) - Converts written text into spoken audio.
* Speech Translation - Translates spoken words into another language.
In this case, the session is transcribed into subtitles in the same language, not translated or spoken aloud, so the correct feature is Speech Recognition.
Let's review the other options:
* Sentiment Analysis: This belongs to the Text Analytics service under natural language processing (NLP) and is used to determine the emotional tone of text, not to convert speech to text.
* Speech Synthesis: Converts text into audible speech (Text-to-Speech), the reverse of what is happening in this scenario.
* Translation: Converts spoken or written words from one language to another. Here, no translation is mentioned-only transcription.
Therefore, the described process-turning live spoken language into readable subtitles-is an example of Speech Recognition, a speech-to-text AI capability provided by Azure Cognitive Services.
Final answer: Speech recognition
Reference:Microsoft Learn - Identify speech capabilities of Azure AI services (AI-900 Learning Path)

According to the Microsoft Azure AI Fundamentals (AI-900) official study guide and Microsoft Learn module "Explore computer vision", Optical Character Recognition (OCR) is a form of computer vision that enables a system to detect and extract printed or handwritten text from images or documents. OCR is particularly useful in scenarios where the goal is to digitize textual information from physical documents, such as receipts, invoices, or travel expense forms - exactly as described in this question.
In the given scenario, employees need a mobile application that allows them to scan and store expenses while traveling. The process involves taking photos of receipts that contain printed text, such as vendor names, totals, dates, and item descriptions. The OCR technology automatically detects the text areas within the image and converts them into machine-readable and searchable data that can be stored in a database or processed further for expense management.
Microsoft's Azure Cognitive Services include the Computer Vision API and the Form Recognizer service, both of which use OCR technology. The Form Recognizer builds upon OCR by adding intelligent document understanding, enabling it to extract structured data from expense receipts automatically.
Other answer options are incorrect for the following reasons:
* A. Semantic segmentation assigns labels to every pixel in an image, typically used in autonomous driving or medical imaging, not for text extraction.
* B. Image classification identifies the overall category of an image (e.g., "This is a receipt"), but it does not extract the textual content.
* C. Object detection identifies and locates objects in an image with bounding boxes but is not used for text reading or conversion.
Therefore, based on the official AI-900 training and Microsoft Learn content, the correct answer is D. Optical Character Recognition (OCR) - the technology that enables extracting textual information from scanned expense receipts.