在基于双麦克风的声学场景分析中，当双麦克风间距变小并且多个声源不断运动时，传统方法对于多个运动声源角度估计的准确性会降低。该文提出了一种适用于近距离双麦克风，对多个运动声源方向进行估计的方法。该方法提出利用改进的Gammatone滤波器对混合声源样本进行时频分析，通过提取与声源方向相对应的过零点时间差（zero-crossing time difference，ZCTD）的特征信息，利用Gauss函数对ZCTD进行统计分析，最终实现多个运动声源方向的确定。对于不同性质、不同速度、不同轨迹的多个运动声源，测试结果表明：该方法对其角度的估计准确性较好。
Angle measurements using traditional auditory analyse with dual-microphones have limited accuracy when the two microphones are close together and the sound sources are moving. This paper presents a method for multiple moving sound source localization with two closely spaced microphones. The method uses an improved gammatone filter bank for the time-frequency analysis of the mixed sound samples, extracts the zero-crossing time difference (ZCTD) that reflects the sound source location feature information and finally statistically analyzes the ZCTD with a Gaussian function to localize the sound source. Tests show that this method gives accurate angle estimates for various kinds of sound sources with various moving speeds along various trajection.
[1] Evans S,Mcgettigan C,Agnew Z,et al.Getting the cocktail party started:Masking effects in speech perception[J].Journal of Cognitive Neuroscience,2016,28(3):483-500.
[2] Simon L S R,Andreopoulou A,Katz B F G.Investigation of perceptual interaural time difference evaluation protocols in a binaural context[J].Acta Acustica United with Acustica,2016,102(1):129-140.
[3] Zeng F G,Rebscher S,Harrison W,et al.Cochlear implants:System design,integration,and evaluation[J].IEEE Reviews in Biomedical Engineering,2008,1(1):115-142.
[4] Gygi B,Hall D A.Background sounds and hearing-aid users:A scoping review[J].International Journal of Audiology,2016,55(1):1-10.
[5] Chen Y,Qin G.Real-time spectrum estimation-based dual-channel speech-enhancement algorithm for cochlear implant[J].Biomedical Engineering Online,2012,11(10):1-22.
[6] Chen Y,Qin G.Broadband beamforming compensation algorithm in CI front-end acquisition[J].Biomedical Engineering Online,2013,12(1):1-20.
[7] Nicoleta R,Deliang W,Brown G J.Speech segregation based on sound localization[J].Journal of the Acoustical Society of America,2003,114(4):2236-2252.
[8] Cui W,Cao Z,Wei J.Dual-microphone source location method in 2-D space[C]//IEEE International Conference on Acoustics,Speech&Signal Processing.Toulouse:IEEE Press,2006:845-848.
[9] Chen Y,Qin G.Real-time spectrum estimation-based dual-channel speech-enhancement algorithm for cochlear implant[J].Biomedical Engineering Online,2012,11(10):2861-2861.
[10] Jourjine A,Rickard S,Yilmaz O.Blind separation of disjoint orthogonal signals:Demixing N sources from 2 mixtures[C]//IEEE International Conference on Acoustics,Speech&Signal Processing.Istanbul:IEEE Press,2000:2985-2988.
[11] Kim Y I,Kil R M.Estimation of interaural time differences based on zero-crossings in noisy multisource environments[J].IEEE Transactions on Audio Speech&Language Processing,2007,15(2):734-743.
[12] 李冰,夏秀渝,申庆超,等.基于过零点双耳时差的运动声源定位[J].计算机工程与应用,2012,48(9):127-130.LI Bing,XIA Xiuyu,SHEN Qingchao,et al.Moving sound localization based on zero-crossing points interaural time differences[J].Computer Engineering and Applications,2012,48(9):127-130.(in Chinese)
[13] 陈世雄,宫琴,金慧君.用Gammatone滤波器组仿真人耳基底膜的特性[J].清华大学学报(自然科学版),2008,48(6):1044-1048.CHEN Shixiong,GONG Qin,JIN Huijun.Gammatone filter bank to simulate the characteristics of the human basilar membrane[J].J Tsinghua Univ (Sci&Tech),2008,48(6):1044-1048.(in Chinese)
[14] Wang D L,Brown G J.Computational Auditory Scene Analysis:Principles,Algorithms,and Applications[M].Piscataway:IEEE Press,1993:15-17.
