A NOISY SPEECH ENHANCEMENT METHOD USING DETECTING METHOD USING A COMPLEX LAPLACIAN PROBABILITY DENSITY FUNCTION

摘要

It is arranged to obtain GSAP (global speech absence probability) using a sound improved method of a complicated Laplace operator probability density function, prediction noise spectrum simultaneously modifies noise elimination income by using GSAP, distribution based on specific sound and noise signals follow complicated Laplace operator probability density function it is assumed that be increased sharply the sound quality with sound mix. It is that distribution based on specific sound and noise signals follows a complicated Laplace operator probability density function it is assumed that a GSAP (global speech absence probability) is obtained. One noise spectrum is predicted by using GSAP. By using noise spectrum and GSAP, noise is eliminated income and is modified. Income is eliminated by using noise spectrum and noise, the tone quality improving with sound mix is to pass through.