针对语音信号中的噪声抑制问题,提出一种基于卷积神经网络(Convolutional Neural Networks,CNN)的噪声抑制优化算法.首先,探讨基于CNN的噪声抑制框架,并研究基于L1正则化的优化方法.利用Noisy Speech Database进行实验测试,比较传统CNN和文章提出的方法在不同噪声环境下的去噪效果.实验结果表明,文章提出方法的信噪比(Signal-to-Noise Ratio,SNR)和均方根误差(Root Mean Square Error,RMSE)均优于传统CNN.
Optimization of Noise Suppression Algorithm Based on CNN
This article proposes an optimized algorithm for noise suppression in speech signals based on Convolutional Neural Networks (CNN). Firstly, explore the noise suppression framework based on CNN and study the optimization method based on L1 regularization. Secondly, experimental tests were conducted using the Noise Speech Database to compare the denoising effects of traditional CNN and the proposed method in different noise environments. The experimental results show that the proposed method has better Signal to Noise Ratio (SNR) and Root Mean Square Error (RMSE) than traditional CNN.
NETL
NSTL
万方数据
