文档介绍：基于非负矩阵分解的单通道语音分离方法研究
摘要
这篇论文首先简单地介绍了一下非负矩阵分解和单通道语音分离。由于数据的非负性的要求,让其更能符合人类的认识思维习惯。非负矩阵分解具有速度快和占用存储空间小等优点,从而使得其得到了广泛的应用,其中包括语音分离。如果能找到一个快速的语音分离算法,它将会推动人工智能、自动音乐复制、声音特效、对象译码等的应用和研究。最后,本文展示了一个用于分离在单个通道记录中的两个讲话者的方法。这个分离实验是在一个代表语音最佳的低维特征空间中实施。对于每个讲话者,一个过完备基是采用稀疏非负矩阵分解评估的,同时语音混合物是通过将其映射到两个讲话者的联合基中分离的。我们采用语音分离竞赛数据集中的单词识别率来评估这个方法。
关键词:非负矩阵分解;单通道;语音分离;语音分离竞赛
Investigating Single-channel speech Source Separation Methods Based on Non-Negative Matrix
Factorization
Abstract
This paper introduce briefly about non-negative matrix factorization (NMF) and Single channel speech separation at first. Because of the requirement of non-negative,makes the NMF fits to the human habit of NMF has the advantage of fast speed and low storage etc.,so,it attain a wide range of applications,which include the separation of we can find a high-effective speech separation algorithm,it will promote the application and research of artificial intelligence ,automatic music transcription, object coding, special sound effects and so ,this paper describes a method for separating two speakers in a single channel recording. The separation is performed in a low dimensional feature space optimized to represent speech. For each speaker, an plete basis is estimated using sparse non-negative matrix factorization, and a mixture is separated by mapping the mixture onto the joint bases of the two speakers. The method is evaluated in terms of word recognition rate on the speech separation challenge data set.
Keywords: Non-negative Matrix Factorization;Single-channel;Speech Separation;speech separation challenge
目录
摘要 I
Abstract II
第一章绪论 1
选题依据 1
国内外研究状况 1
2
论文主要成果 3
论文结构 3
第二章语音分离及NMF背景知识 5
语音分离 5
非负矩阵分解(NMF) 6
第三章 NMF算法分析 8
8
8
迭代函数 8
局部非负矩阵分解(LNMF) 9
目标函数 9
9
非负稀疏编码(NNSC) 10
10
10
稀疏非负矩阵分解(SNMF) 10
10
11
第四章实验 12