基于新型组合算法的语音降噪系统

在研究VSLMS和子空间降噪原理的基础上，分析了两种算法的优缺点，提出了一种组合降噪算法，并进行了硬件实现。组合降噪算法试图寻找一个语音信号的分解点，利用子空间算法分离出带噪信号，作为自适应算法的输入，最后由VSLMS算法对带噪信号进行增强处理。硬件实现在该组合降噪算法的基础上，利用SOPC技术在FPGA芯片中搭建嵌入式语音降噪系统。实验结果表明，该方法能够优于传统降噪算法，使用硬件结构加快了算法的收敛速度。

Speech Noise Reduction System Based on A New Combined Algorithm

This paper proposes and hardware implements a combined filtering algorithm based on the principles of VSLMS and the subspace filtering algorithm. The subspace algorithm tries to find a voice signal decomposition point and separates the signal from the noise, so that the signal can be used as the input of VSLMS. The signal is then enhanced by VSLMS. The SOPC technology is employed to build the embedded voice noise reduction system on the basis of software algorithm. The experiment results show that the method is superior in convergence speed to the traditional noise reduction algorithm by its use of the hardware architecture.