Question

You are building a voice-enabled agent for a pharmacy that reads medication instructions aloud using Azure AI Speech text-to-speech. Testers report two problems: drug names like 'metoprolol' are mispronounced, and dosage phrases such as 'take 1 tablet twice daily' are spoken too quickly for elderly users to follow. You must fix both issues without training a custom neural voice. What should you do?

Accepted Answer

Use SSML with a element to specify the correct pronunciation of drug names and a element to slow the speaking rate for dosage phrases

Answer

Switch the synthesis output format to a higher-bitrate audio codec and increase the audio sample rate

Answer

Send the text through Azure Translator before synthesis to normalize spelling, then synthesize at the default rate

Answer

Train a Custom Neural Voice using recordings of a pharmacist so the model learns the drug names and slower pacing