一种语音与背景乐音信号盲分离优化算法

在盲分离算法中,许多智能优化算法被应用,以克服独立性准则函数的优化进入局部最优位置,但这些优化算法的性能依赖控制参数的选择。因此,提出利用一种单参数的纯随机搜索的单形进化优化算法(Surface-Simplex Swarm Evolution,SSSE),克服算法参数对优化算法性能的影响,提高盲分离算法的应用有效性,并将该改进盲分离算法应用于语音与背景乐音信号的盲分离。实验中,以四阶累积量作为独立分量分析(Independent Component Analysis,ICA)中的准则函数。实验结果表明,该改进算法有效分离出语音成分与背景乐音成分,而且在稳定性和分离效果方面具有较好的性能。

Optimization Algorithm for Blind Separation of Speech and Background Music Signals

In the blind separation algorithm,many intelligent optimization algorithms are applied to overcoming the optimization of the independence criterion function into the local optimal position,but the performance of these optimization algorithms usually depends on the choice of control parameters.Therefore,a SSSE(Surface-Simplex Swarm Evolution)algorithm is proposed so as to overcome the influence of algorithm parameters on the performance of the optimization algorithm and improve the application effectiveness of the blind separation algorithm.The modified blind separation algorithm is applied to the blind separation of speech and background music signals.In the experiment,the fourth-order cumulant is used as the criterion function in ICA(Independent Component Analysis).The experimental results show that the modified algorithm can effectively separate the speech and background music components,and has better performance in terms of stability and separation effect.