First, VAD is done to noisy speech, then spectral subtraction based on the perceptual mask threshold follows to achieve noise reduction.

• 这种新算法源于VAD技术及谱减法,它先对含噪声信号进行VAD处理,而后是基于感知掩蔽阈值的谱减法运算,降低噪声,获得清晰的增强信号。