speechToText needs to specify binary mode

[BruceWilliams]

in the implementation of of speechToText, it currently reads from the audio file like so:

filecontents = File.read(file)

This works on Ruby 1.8, which assumes binary mode. It fails on Ruby 1.9 and later, because you have to explicitly specify binary mode:

filecontents = File.read(file, :mode => 'rb')

The symptom of the failure is that it only gets the first 8000 or so bytes of the audio file, which is enough to get processed by the back-end, but generally not enough to be properly recognized and converted to text.