针对语音信号的盲提取问题,以提出的二阶特征窗盲分离算法和模式识别算法为基础,设计实现了以TMS320VC5509 DSP为核心的音频信号采集和盲分离系统。详细介绍了系统结构及软硬件实现,并对3路音频信号的DMA同步采集进行了重点分析。该系统结构简单、扩展方便、实时性好,实验结果表明对瞬时混合信号有良好的分离效果。
A speech signal acquisition and blind source separation system is designed based on the second-order characteristic window blind source separation algorithm and pattern recognition algorithm is proposed to solve the blind speech extraction problem. The system structure and its software and hardware design are discussed in detail. The DMA synchronal acquisition method of three channels audio signal is analyzed. The proposed system structure is simple, and it can be expanded easily and can be implemented in real-time. The experimental results demonstrate that the instantaneous mixture speech signals can be separated effectively by the proposed system.