SPEECH ENHANCEMENT USING THREE-STEP-DECISION GAIN FACTOR WITH OPTIMAL SMOOTHING

摘要

The spectral estimate of speech plays a major role in computing the noise masking threshold when simultaneous masking properties are employed for speech enhancement. Traditional methods using the power-spectral-subtraction method to roughly estimate the speech spectra can provide acceptable performances, however the performance can be further improved. In this study, we aim at finding a better speech spectral estimate through a modified two-step-decision-directed method. In turn, this speech spectral estimate is employed to compute the noise masking threshold, enabling the Virag speech enhancement algorithm to be improved. Experimental results show that the proposed method outperforms the two-step-decision-directed and the Virag algorithms in most conditions.