azure - Bing Voice Recognition API Output - Number vs Text -
i'm seeing odd behavior voice recognition api return words , return numbers. example, given audio input of "fifteen dollars , 7 cents" return "fifteen dollars , 7 cents" , others "$15.07". there way ensure 1 or other returned? or, need parse , convert words numbers on own.
thanks!
i looked similar issues, , found bingstt api returns both versions in different fields. tried example example code here: https://github.com/microsoft/cognitive-speech-stt-javascript , got response:
[{ "lexical": "fifteen dollars , 7 cents", "display": "$15.07.", "inversenormalization": null, "maskedinversenormalization": null, "transcript": "$15.07.", "confidence": 0.9474185 }]
so guess choice yours field use, lexical or display. hope helps.
Comments
Post a Comment