Skip to content

What is the latency of calling Azure TTS

ForrestGumb edited this page Aug 22, 2024 · 6 revisions

Latency is very critical metrics for a production speech application. It is also a key metrics in Azure TTS production development. We try best to optimize the run-time and models for minimum latency while keeping the best quality.

Latency test needs to be conducted in a specific setting. It is highly depending on where the client machine and Azure DC are.

To conduct a latency test, here are a few check list

  • choose an Azure region closing to the service calling into Azure TTS.
  • make sure to reuse http connection or synthesizer object in SDK which can reduce re-connection latencies.
  • try multiple times to calculate average, percentile latency numbers.

We have multiple partners in 1st and 3rd party chosen to use Azure TTS in real time applications. See more on - Where can I see the demo or app using Azure TTS service?