CertPrep Coach

Text and speech solutions

Build apps that analyze text, use Azure Speech for speech to text and text to speech, and answer spoken prompts with a multimodal model.

Two ways to build a voice experience A pipeline turns speech into text, processes the text, and turns the reply back into speech with Azure Speech. Alternatively, a deployed multimodal model can take the spoken prompt directly and answer it. Pipeline: Azure Speech + language processing Single multimodal model Spoken input microphone or recording Speech to text transcribe audio Process text text analysis or a language model Text to speech humanlike voice Reply spoken Spoken prompt audio input Deployed multimodal model understands audio directly Answer text, or audio if the model supports it
A pipeline turns speech into text, processes the text, and turns the reply back into speech with Azure Speech. Alternatively, a deployed multimodal model can take the spoken prompt directly and answer it.

What Microsoft lists for this area

  • Build a lightweight application that includes text analysis
  • Respond to spoken prompts by using a deployed multimodal model
  • Build a lightweight application by using Azure Speech in Foundry Tools

What this area tests

  • building a lightweight app that includes text analysis
  • responding to spoken prompts with a deployed multimodal model
  • building a lightweight app with Azure Speech in Foundry Tools

Text analysis in an app

There are two common ways to analyze text in an app:

  1. Call a ready-made text analysis feature, such as sentiment analysis or entity recognition in Azure Language. You get consistent, structured results without writing prompts.
  2. Prompt a language model to do the analysis, for example “Classify this review as positive, negative or neutral and list the products mentioned.” This is flexible, and you can ask for output in JSON so your app can read it.

Azure Speech

Azure Speech in Foundry Tools provides speech to text, text to speech and other capabilities through a Microsoft Foundry resource.

  • Speech to text converts audio into text, either live from a microphone or from recorded files.
  • Text to speech converts input text into humanlike synthesized speech, with a choice of voices.

A classic voice app is a pipeline: speech to text, then process the text (with text analysis or a language model), then text to speech for the reply.

Spoken prompts with a multimodal model

Some deployed models are multimodal: they accept audio as input, and some can also reply with audio. With one of these, you can skip the separate transcription step and send the spoken prompt straight to the model. This is simpler to build, but you depend on the model’s audio support.

Choosing between them

NeedUse
Accurate transcripts of calls or meetingsAzure Speech, speech to text
A natural voice reading text aloudAzure Speech, text to speech
Consistent sentiment or entity resultsA text analysis feature
Flexible analysis described in plain languageA language model with a prompt
A voice assistant that understands speech directlyA multimodal model that accepts audio

Practice Text and speech solutions

19 practice questions with explanations for this area.

Practice this area Or read 12 sample questions with answers

Sources

Updated October 25, 2026. Microsoft changes products often; check the official study guide for the latest outline.