Ordinary copper telephone service has enough bandwidth to transmit a full range of audio signals, in fact more than the voice can produce or the ear can hear. Modems exploit this fact.
But cell phones employ lossy compression, and if I understand correctly, one of the "first things to go" are the higher frequency noises. Why transmit what the ear can't hear?
If each of his minutes spent transferring nothing gets him 2mb I suppose he is 1.58mb up if he can maximise his transfer.