Recently we've been seeing increased error output when performing speech synthesis using the Node.JS SDK. This seems to happen after a number of transactions (30-60), but not every time. The errors appear mixed, with the first starting to appear before the second error starts appearing. The same content has been run through synthesis without errors before. We're on a Standard (S0) subscription, so the higher rate limits should apply. Our resource region is set to eastus.
Status(StatusCode="NotFound", Detail="service does not exist: service endpoint xds:///bg-bg-combogro websocket error code: 1007
Status(StatusCode="ResourceExhausted", Detail="no tokens available for passthrough") websocket error code: 1013