chia-hsin yang | National Taiwan Ocean University (original) (raw)

chia-hsin yang

Uploads

Papers by chia-hsin yang

Research paper thumbnail of Estimation of Sound Source Number and Directions under a Multisource Reverberant Environment

EURASIP Journal on Advances in Signal Processing, 2010

Sound source localization is an important feature in robot audition. This work proposes a sound s... more Sound source localization is an important feature in robot audition. This work proposes a sound source number and directions estimation method under a multisource reverberant environment. An eigenstructure-based generalized cross-correlation method is proposed to estimate time delay among microphones. A source is considered as a candidate if the corresponding time delay combination among microphones gives reasonable sound speed estimation. Under reverberation, some candidates might be spurious but their direction estimations are not consistent for consecutive data frames. Therefore, an adaptive K-means++ algorithm is proposed to cluster the accumulated results from the sound speed selection mechanism. Experimental results demonstrate the performance of the proposed algorithm in a real room. K k=1

Research paper thumbnail of Estimation of Sound Source Number and Directions under a Multisource Reverberant Environment

EURASIP Journal on Advances in Signal Processing, 2010

Sound source localization is an important feature in robot audition. This work proposes a sound s... more Sound source localization is an important feature in robot audition. This work proposes a sound source number and directions estimation method under a multisource reverberant environment. An eigenstructure-based generalized cross-correlation method is proposed to estimate time delay among microphones. A source is considered as a candidate if the corresponding time delay combination among microphones gives reasonable sound speed estimation. Under reverberation, some candidates might be spurious but their direction estimations are not consistent for consecutive data frames. Therefore, an adaptive K-means++ algorithm is proposed to cluster the accumulated results from the sound speed selection mechanism. Experimental results demonstrate the performance of the proposed algorithm in a real room. K k=1

Log In