压缩感知下的稀疏表示语声恢复模型与算法

发布时间：2018-02-08 12:15

本文关键词： 语声恢复压缩感知稀疏表示　出处：《信号处理》2014年08期 　论文类型：期刊论文

【摘要】：本文讨论的语声信息恢复旨在提高带噪语声的可懂度。通过类比听觉掩蔽与视觉闭塞,在基于稀疏表示的图像去噪思想启发下,本文提出了基于压缩感知理论的稀疏表示语声恢复模型、数学表达式以及算法。与传统的语声增强算法不同,本文模型与算法的特点在于具备有效消除全局噪声干扰和恢复局部被噪声掩蔽的语声成分的双重能力,有效提高了处理后语声的可懂度。仿真实验和客观语声质量测度验证了提出的模型与算法的可行性、有效性以及优越性。
[Abstract]:The speech information restoration discussed in this paper aims to improve the intelligibility of noisy speech. By analogy between auditory masking and visual blocking, the image denoising idea based on sparse representation is used. This paper presents a sparse representation speech restoration model, mathematical expressions and algorithms based on compressed perception theory, which are different from the traditional speech enhancement algorithms. The characteristic of the model and algorithm is that it can effectively eliminate the global noise interference and recover the local noise masked sound components. The intelligibility of the processed speech is improved effectively, and the feasibility, validity and superiority of the proposed model and algorithm are verified by simulation experiments and objective speech quality measurement.
【作者单位】：中国科学院声学研究所;
【分类号】：TN912.35

【相似文献】