Pipelined architecture for neural-network-based speech recognition