Study Azure AI-900 Language Services: key concepts, common traps, and exam decision cues.
After you identify the workload, AI-900 may ask which Azure service family fits it. The key boundary is simple: Azure AI Language is text-first, while Azure AI Speech is audio-first.
| Requirement | Strongest first fit |
|---|---|
| sentiment, entities, key phrases, or text understanding | Azure AI Language |
| speech-to-text or text-to-speech | Azure AI Speech |
| audio-first translation or spoken-language interaction | Azure AI Speech |
| written-text analysis | Azure AI Language |
Use this sequence:
This rule solves many AI-900 distractors before you read every answer choice.
| Trap | Better rule |
|---|---|
| choosing Speech because a scenario mentions language | Speech is for audio-first tasks, not for every language task |
| choosing Language because a transcript appears later in the workflow | if the system must create the transcript from audio, Speech is the stronger first fit |
| treating text analysis as generative AI by default | analyzing existing text is not the same as generating new text |