the following works in curl as long as the file sent is not too long... under 5 seconds.
curl -X POST -H "Content-Type: audio/x-flac; rate=16000" \ -T seg_1.flac "https://www.google.com/speech-api/v1/recognize? \ xjerr=1&client=speech2text&maxresults=1&lang=en-US&key=...48593"
{"status":0,"id":"","hypotheses":[{"utterance":"now it was the favorite pastime","confidence":0.95148802}]}
So, encode to speechX or flac
include a parm with your sample rate from the recording
include your key
keep the file short in duration ( you will have to split files prior to API access )