This might be helpful, @cameron created a skill that uses Google’s DialogFlow to do what you’re probably attempting yourself.
In essence the simplest thing to do is to send the audio file as a base64 string (Google Cloud Services accepts audio in that format). Check out
function ProcessAudioFile(data) which uses the
misty.SendExternalRequest() call to see how Cameron accomplishes this!
Out of curiosity, are you writing your own ‘intent detection’ method or are you utilizing Google/Watson/Cognitive services?
Let me know if this helps!