Taking into account the masking properties of human auditory system and the uncertainty of the speech pres-ence

the novel algorithm modifies the log-spectral amplitude estimation and the speech absence probability（SAP） pa-rameter.Then

the smoothing parameters in noise estimation and prior SNR estimation are adaptively adjusted to modi-fied SAP parameter which reflects the degree of stationary of the measured signal.The change of smoothing parameters allows for automatic adaptation with various noisy environments and obtains the best trade-off.Experimental results demonstrate that the proposed algorithm has better performance of speech articulation without any perceptional musical-ity

compared to another modified spectral subtraction algorithms under the same level of noise reduction

and this supe-riority is more significant at very low SNR.