基于DSP的小阵列语音增强算法的研究与实现

以TI公司TMS320C5535 DSP和二元麦克风小阵列为基础,提出一种行之有效的语音增强算法。通过分析在小阵列上常用的广义旁瓣相消器算法和相干滤波器算法的优缺点,将时变平滑因子引入到相干滤波器中,并将相干滤波器置于广义旁瓣相消器的固定波束形成支路。在得出一种有效算法的同时也给出了其在DSP上的实时实现步骤。仿真与实际结果表明,在小阵列上该方法明显优于单独使用广义旁瓣相消器算法和相干滤波器算法,可以有效抑制噪声干扰,提高语音质量。

Speech Enhancement Method based on the Combination of Coherence Filter and Generalized Sidelobe Canceler for Small Microphone Arrays

An effective speech enhancement algorithm is proposed based on TI"s TMS320C5535 DSP and dual small microphone array. By analyzing the advantages and disadvantages of generalized sidelobe canceller algorithm and coherent filter algorithm which is often used for small arrays, a time-varying smooth factor is introduced into the coherent filter in the paper, and then the coherent filter is placed on fixed beamforming branch of generalized sidelobe canceller. Finally, an efficient algorithm is derived and its real-time implementation steps on the DSP are also presented in this paper. Simulations and practical tests show that the proposed algorithm is superior to separate small array generalized sidelobe canceler algorithms and coherence filter algorithm.