提出一种基于交替方向乘子法的(Alternating Direction Method of Multipliers,ADMM)稀疏非负矩阵分解语音增强算法,该算法既能克服经典非负矩阵分解(Nonnegative Matrix Factorization,NMF)语音增强算法存在收敛速度慢、易陷入局部最...提出一种基于交替方向乘子法的(Alternating Direction Method of Multipliers,ADMM)稀疏非负矩阵分解语音增强算法,该算法既能克服经典非负矩阵分解(Nonnegative Matrix Factorization,NMF)语音增强算法存在收敛速度慢、易陷入局部最优等问题,也能发挥ADMM分解矩阵具有的强稀疏性。算法分为训练和增强两个阶段:训练时,采用基于ADMM非负矩阵分解算法对噪声频谱进行训练,提取噪声字典,保存其作为增强阶段的先验信息;增强时,通过稀疏非负矩阵分解算法,从带噪语音频谱中对语音字典和语音编码进行估计,重构原始干净的语音,实现语音增强。实验表明,该算法速度更快,增强后语音的失真更小,尤其在瞬时噪声环境下效果显著。展开更多
The paper tests the impact of the evolution of China’s investment and sourcing mechanism on its economic growth with a structural model in which aggregate output and investment are treated as endogenous variables.Our...The paper tests the impact of the evolution of China’s investment and sourcing mechanism on its economic growth with a structural model in which aggregate output and investment are treated as endogenous variables.Our estimates using the methods of indirect least squares show a strong correlation between output and investment supporting the hypothesis that China’s growth are almost fully attributable to its capital investment.However,the results also indicate that it is the institutional reforms regarding investment ownership diversification, market orientation and local government participation that supported China’s lasting growth in the past two decades.Our empirical tests also produce a negative correlation between China’s output growth and its level implicating that growth convergence is applicable to China.展开更多
文摘提出一种基于交替方向乘子法的(Alternating Direction Method of Multipliers,ADMM)稀疏非负矩阵分解语音增强算法,该算法既能克服经典非负矩阵分解(Nonnegative Matrix Factorization,NMF)语音增强算法存在收敛速度慢、易陷入局部最优等问题,也能发挥ADMM分解矩阵具有的强稀疏性。算法分为训练和增强两个阶段:训练时,采用基于ADMM非负矩阵分解算法对噪声频谱进行训练,提取噪声字典,保存其作为增强阶段的先验信息;增强时,通过稀疏非负矩阵分解算法,从带噪语音频谱中对语音字典和语音编码进行估计,重构原始干净的语音,实现语音增强。实验表明,该算法速度更快,增强后语音的失真更小,尤其在瞬时噪声环境下效果显著。
文摘The paper tests the impact of the evolution of China’s investment and sourcing mechanism on its economic growth with a structural model in which aggregate output and investment are treated as endogenous variables.Our estimates using the methods of indirect least squares show a strong correlation between output and investment supporting the hypothesis that China’s growth are almost fully attributable to its capital investment.However,the results also indicate that it is the institutional reforms regarding investment ownership diversification, market orientation and local government participation that supported China’s lasting growth in the past two decades.Our empirical tests also produce a negative correlation between China’s output growth and its level implicating that growth convergence is applicable to China.