I have couple of question related to mobile network generations:
They use maximum bandwidth to transmit analogue voice information. Since available bandwidth is limited and there is a limit on the bandwidth needed to make the voice recognizable on the other side, they can't go below this fixed bandwidth.
It refers to the frequency span of the carrier. If it is narrow band, then usually it's just a few kilohertzes, which makes available bandwidth low (this also depends on the kind of modulation used).
They don't need the maximum available bandwidth for voice communications, since for an acceptable voice quality we need to transmit signals that encode just a few kilohertzes of analogue information.