论文部分内容阅读
为了提高噪声和混响条件下分布式传声器阵列进行声源定位的性能,提出一种利用空间稀疏性和压缩感知原理的声源三维定位方法。该方法首先通过两次离散余弦变换方式提取出声音信号特征,并用该特征来构建稀疏定位模型,以便能够综合利用语音信号的短时和长时特性,同时降低模型维数;然后利用在线字典学习技术动态调整字典,克服稀疏模型与实际信号之间的失配问题,增强稀疏定位模型的鲁棒性;进而提出一种改进的平滑l_0范数稀疏重构算法来进行声源位置解算,以提高低信噪比条件下的重构精度。仿真结果表明该方法不仅可以实现多目标定位,而且具有较强的抗噪声和抗混响能力.
In order to improve the localization performance of the distributed microphone array under noise and reverberation conditions, a three-dimensional localization method of sound source using spatial sparsity and compressive sensing is proposed. This method first extracts the features of the sound signal by twice discrete cosine transform, and constructs the sparse localization model by using this feature, in order to make full use of the short-term and long-time characteristics of the speech signal and reduce the dimension of the model. Then, Adjust the dictionary dynamically to overcome the mismatch between the sparse model and the actual signal and enhance the robustness of the sparse localization model. Then, an improved smooth l_0 norm sparse reconstruction algorithm is proposed to solve the problem of Improve the reconstruction accuracy under low signal-to-noise ratio. Simulation results show that this method not only can achieve multi-target localization, but also has strong anti-noise and anti-reverberation capability.