Parallel and distributed encoding of speech across human auditory cortex