Abstracts Submitted: 464
Number of Users: 528
View Abstracts Submitted
Back to home Page
Estimating the location of the sound source in the room is useful for grasping the situation in the room. The Generalized cross correlation phase transform (GCC-PHAT) has been proposed to estimate a location of a sound source by using the arrival time difference of sound wave acquired by a microphone array. However, when a general microphone is used, it is difficult to estimate the location of an indoor sound-source from outside because the sound wave is blocked and does not arrive at the microphone. In this research, we propose a method of sound source localization using a laser microphone to estimate the location of an indoor sound-source from outside. This micro-phone irradiates a laser light to an object vibrating by sound waves and measures vibration to acquire an acoustic signal. Thus, even if a microphone is not installed near the sound source, it is possible to acquire the desired signal by irradiating the laser light to the object near the sound source. In this research, we used a laser Doppler vibrometer (LDV) as a laser microphone. In this research, we used a LDV-array to simultaneously irradiate multiple vibrating objects with the laser light. In the acquired acoustic signals, the arrival time difference of sound waves occurs depending on the distance between each vibrating object and the sound source. When the distance between the vibrating objects is known, the location of a sound source can be estimated from the arrival time difference of sound waves. However, the acquired acoustic signal by the LDV includes noise occurred by vibration of the LDV body itself and the estimation error of the arrival time difference increases. Thus, in this research, we examined the sound source localization method which carried out the two processes of voice activity detection and frequency selection of input signal in the proposed method. We considered that estimation errors can be reduced by extracting only the tar-get voice of the input signal using these two processes. Moreover, we conducted an evaluation experiment to evaluate the performance of the proposed method. As a result of the experiment, we confirmed that the proposed method can estimate the location of an indoor sound-source from outside with the same estimation error as in the case of estimating the location of a sound source by placing the general microphone at the location of the vibrating object. We also confirmed that estimation errors can be reduced by using two processes of voiced activity detection and frequency selection.
© Copyright 2017 All Rights Reserved