基于正弦加噪声模型的说话人转换方法.pdf
Sine plus noise model-based speaker conversion method. Pdf (2012-08-19, PDF, 367KB, 下载5次)
《模式分类》(原书第2版)已被卡内基-梅隆、哈佛、斯坦福、剑桥等120多所大学采用为教材。本书作为流行和经典的教材和专业参考书，主要面向电子工程、计算机科学、数学和统计学、媒体处理、模式识别、计算机视觉、人工智能和认知科学等领域的研究生和相关领域的科技人员。开发和研究模式识别系统的实践者，无论其应用涉及语音识别、字符识别、图像处理还是信号分析，常会遇到需要从大量令人迷惑的技术中做出选择的难题。这本独一无二的教材及专业参考书，为你准备了充足的资料和信息，供你选择最适合的技术。
"Pattern Classification" (the original book version 2) has been Carnegie- Mellon, Harvard, Stanford and Cambridge, more than 120 universities use for teaching purposes. Book as popular and classic textbooks and professional reference books, mainly for electrical engineering, computer science, mathematics and statistics, media processing, pattern recognition, computer vision, artificial intelligence and cognitive sciences graduate student and related fields IT staff.
(2011-08-29, PDF, 7483KB, 下载19次)
贝叶斯压缩感知在语音识别上的一种应用方法
Bayesian compressed sensing in a speech recognition application method on (2011-06-15, PDF, 103KB, 下载32次)
噪声环境下的端点检测在语音信号分析和识别中占有重要地位。文中将分形理论中的分形记盒维数应用到端点检测算法中，采用了基于分形记盒维数与短时能零比相结合的端点检测算法，以分形记盒维数为主要判决条件，并在判决门限的设定上采用了自适应机制。
Noise environment endpoint detection in speech signal analysis and identification play an important role. Wen will be fractal theory of fractal box dimension in mind applied to the endpoint detection algorithm, using case based on fractal dimension and short-term in mind can be nil to a combination of endpoint detection algorithm to the fractal box dimension in mind as the main sentence conditions, and decision threshold setting on the use of adaptive mechanisms. (2010-02-12, PDF, 871KB, 下载28次)
基于短时谱估计的语音增强研究，包括谱减法，维纳滤波和MMSE的介绍，很详细。
Based on the short-time spectral estimation of speech enhancement research, including spectral subtraction, Wiener filter and the MMSE introduction in great detail. (2009-06-04, PDF, 446KB, 下载81次)