一种采用振荡神经网络的CASA计算模型语音分离算法

基于听觉现象分析（CASA）模型的基本原理，在仅有单通道输入混合语音信号时，采用振荡器神经网络，提出了一种CASA改进模型语音分离算法结构，文中利用一个实例说明了新算法的具体实现步骤，讨论了新算法机构中语音听觉外围处理部分和分割神经网络处理部分，通过上述两个部分的处理可以将输入混合语音信号在时频域上分割为若干有听觉感知意义的语音听觉感知成分分段Segments，以便于新算法后续处理分部中语音Segments的聚为和分离重构输出处理，最终完成语音分离任务。

Computational Auditory Scene Analysis Speech Separation Algorithm Using Oscillator Neural Network

Based on the principle of computational auditory scene analysis (CASA), a CASA speech separation algorithm was proposed by using oscillator neural network. One example is used to explain the processing and the coefficients in the proposed algorithm in detail. The first part of the proposed algorithms includes the auditory periphery processing and the processing of the segmentation oscillator neural network. The speech auditory perceptive segments can be given by the processing of this part in the proposed algorithm, and they will be used for the clustering and reconstructing processing in the next part of the proposed algorithm.