error in calling speech_synthesizer.speak_ssml_async(ssml).get()

Ken L 5

when running speech_synthesis_result = speech_synthesizer.speak_ssml_async(ssml).get()

sometime it works, sometimes it shows:

Speech synthesis canceled: CancellationReason.Error

Error details: Connection was closed by the remote host. Error code: 1013. Error details: Status(StatusCode="ResourceExhausted", Detail="no tokens available for passthrough") USP state: TurnStarted. Received audio size: 0 bytes.

Did you set the speech resource key and region values?

what is the problem and how to fix it? The key and region values are all correct.

ryan kilgour 0 Reputation points

2023-04-30T14:11:18.26+00:00

I am also having this problem when generating with SSML with certain voices. It seems to work with some but not others. I have been using en-CA-ClaraNeural for testing in the meantime and this seems to work for me, however it has dramatically limited the number of options I have in my app so I am interested in the solution.
VasaviLankipalle-MSFT 14,416 Reputation points

2023-05-01T20:37:44.0933333+00:00

Hi @Ken L , Thanks for using Microsoft Q&A Platform.

This error can occur if you are making too many requests to the speech synthesis service.

Please try to check the speech resource in the Azure portal. Also, check for the pricing tier of your Speech resource and its request limit.

Could you please share the region of your resource?