Question

A publishing company needs to convert 800 full-length novels (each 60,000+ words) into narrated audio files using Azure AI Speech. The process runs overnight as a scheduled job, and the resulting MP3 files are stored in Azure Blob Storage for later distribution. Latency per request is irrelevant, but the solution must handle very large text inputs efficiently without maintaining a persistent client connection. Which text-to-speech approach should you use?

Accepted Answer

Use the Batch synthesis API, submitting each novel as an asynchronous job that writes output audio to the configured Blob Storage container.

Answer

Use the real-time Speech SDK SynthesizeToStreamAsync method in a loop, chunking each novel into small segments.

Answer

Use the real-time text-to-speech REST endpoint, sending the entire novel text in a single POST request per book.

Answer

Use Speech Studio's live audio preview to render each novel and manually download the resulting files.