first, you're likely capturing stereo audio, so the volume of data sent will be very high. I'd also try to find ways to reduce the volume of audio being sent. IEEE float at 44.1kHz is not very efficient.
Why not start by capturing the audio to a WAV file, and saving a few seconds. This might highlight any mismatches of WaveFormat.
Why not start by capturing the audio to a WAV file, and saving a few seconds. This might highlight any mismatches of WaveFormat.