A robust speech recognit ion system based on discriminative learning of environmental features is

proposed for recognit ion of environmental features in high noise background, and a g radient descent algo􀀂

rithm was adopted for parameters opt imization. Ex periments w ere carried out at different levels of back􀀂

g round noise for SNR, basic accuracy, noise- resistance and system environment adaptability. T he experi􀀂

mental result s show that the system has good recog nition performance in high noisy environments. The sys􀀂

tem can meet different needs of applicat ion.