0

I have below 2 queries:

  1. Can we use VXML based grammars with the Microsoft's cloud speech recognition service ?
  2. Can we use compiled(binary-format) grammars with Microsoft's cloud speech recognition service?

We have integrated C++ SDK provided by Microsoft to use these services. I have tried to find what are the types of grammars supported and could only find something called as PhraseListGrammars(https://learn.microsoft.com/en-us/python/api/azure-cognitiveservices-speech/azure.cognitiveservices.speech.phraselistgrammar?view=azure-python).

Please help us if you are know how much microsoft's cloud STT service supports such custom grammars.

Thanks

1 Answers1

0

@aman bhardwaj For the above scenario both the options VXML and complied binaries are not supported. The option that is available for customization or improvement of grammar is to use the custom speech to text option. This allows users to upload test datasets to train custom models to recognize formats that are otherwise not recognized.

For example: The spoken form of a particular abbreviation.

Spoken form of this abbreviation IEEE is "i triple e" which is not recognized by standard speech to text response, a custom model can be trained with pronunciation files to recognize the spoken form and display the correct text. Most of these custom scenarios are applicable to industries where custom terms are used and the cloud STT API should be able to provide correct grammatical response for the speech input provided.

RohitMungi
  • 124
  • 6