InitialSilenceTimeout when using Rest API

Question

I am using the Cognitive service REST API to send chunks of audio (approximately 20s). However, some chunks have a long initial period without a human voice - just silence or the vignette sound).

In these cases, we are receiving InitialSilenceTimeout and no transcriptions in the API response.

Is there a way to tell the REST API to wait longer for the speech to start?

This is the endpoint we are using to perform speech-to-text:

`https://eastus.stt.speech.microsoft.com/speech/recognition/conversation/cognitiveservices/v1?language=pt-BR`  

Thanks in advance,

Rhenan

Answer 1

@Rhenan Bartels I believe you might be seeing a response like below if there is a long silence in your audio file.

{  
    "RecognitionStatus": "InitialSilenceTimeout",  
    "Offset": 50000000,  
    "Duration": 0  
}  

This is because the short audio API does not allow setting the initial silence parameter using REST API. The only ones supported are mentioned here.
You should try to switch to using any of the SDK and set the InitialSilenceTimeout parameter of the speech recognition engine.

If an answer is helpful, please click on or upvote which might help other community members reading this thread.