基于麦克风阵列的混合车辆音频信号分离

Hybrid Vehicle Audio Signal Separation Based on Microphone Array

麦克风阵列接收到的车辆信号一般为几辆车的混合，若对单个车进行处理，必须对混合车辆信号进行分离。波束形成和盲源分离算法都是经典的语音分离算法，在对车辆信号分离中应用较少。比较两种方法在分离车辆音频效果后，提出了将两种算法相结合的方法。通过实验仿真，两种方法的结合实现了各个入射角度的信号分离，解决了车辆距离较近时不能得到较纯净源信号的问题。

The vehicle signal received by the microphone array is usually mixed with several vehicles. If a single vehicle is processed, the mixed vehicle signal must be separated. Beam forming and blind source separation are classical speech separation algorithms, and they have fewer applications in the separation of vehicle signals. After separating the audio effects of the vehicle, the method of combing the two algorithms is proposed. Through ex- periment simulation, the combination of the two methods realizes the separation of signals from each incident angle, and solves the problem that the vehicles are not able to get the pure source signals when they colse to each other.