Microphone array sound source spatial real-time location method

Abstract

The invention provides a microphone array sound source spatial real-time location method. A microphone array is used as a signal acquisition and output device. Sound source spatial candidate positions are preliminarily given through the use of a controllable power response-phase transformation method. Preliminary candidate filtering is carried out through prior knowledge. The controllable power response-phase transformation method is used to calculate the controllable power response output of the candidates. A search boundary is re-determined through improved random region shrinkage. The efficiency of the controllable power response-phase transformation method is improved. Finally, the controllable power response of the remaining candidates is calculated. The maximum position is selected as the final sound source estimation position. According to the invention, the method has the advantages of clear sound source positioning principle and good real-time performance; experiments show that the plane positioning error range of the method can be controlled on the order of centimeter; the performance is superior to the performance of methods in the prior art; the method has high operation speed and robustness, and can be applied to intelligent home, intelligent robots and other scenes which need real-time sound source localization.

Description

technical field [0001] The invention belongs to the technical field of speech, and in particular relates to a method for real-time positioning of a microphone array sound source space. Background technique [0002] Speech processing technology usually needs to master the spatial position information of the sound source to provide assistance for interactive technologies such as voiceprint recognition and speech content recognition. For example, when a user has a voice conversation with an intelligent robot, it is usually necessary to determine the spatial position of the person and turn and approach. Even in the dark, the specific location of the user can be determined only based on the voice, and it will automatically approach the speaker, providing the necessary Service and help; In video conferences, it is generally necessary to automatically adjust the conference camera to face the speaker, and to be able to switch between different speakers. [0003] A common device use...

AI Technical Summary

Problems solved by technology

[0023] The sound source localization method based on high-resolution spectrum estimation draws on the positioning technology in radar arrays, but radar signals are mostly narrow-band signals, compared with the wide frequency band of voice signals, so the effect of this method is not ideal

[0024] Sound source localization is widely used in human-computer interaction technology, but traditional methods cannot achieve high robustness and real-time effects at the same time

[0025] It can be seen that the sound source localization method based on microphone array has been widely studied. Compared with other methods, the localization method based on controllable power response has the advantage of resisting environmental noise and reverberation interference, but its real-time calculation is poor. Difficult to play a role in application scenarios

Images