• Journal of Broadcast Engineering
• /
• v.21 no.2
• /
• pp.169-179
• /
• 2016

Recently, MPEG-H has been standardizing for a multimedia coder in UHDTV (Ultra-High-Definition TV). Thus, the demand for not only channel-based audio contents but also object-based audio contents is more increasing, which results in developing a new technique of converting channel-based audio contents to object-based ones. In this paper, a non-uniform linear microphone array based source separation method is proposed for realizing such conversion. The proposed method first analyzes the arrival time differences of input audio sources to each of the microphones, and the spectral magnitudes of each sound source are estimated at the horizontal directions based on the analyzed time differences. In order to demonstrate the effectiveness of the proposed method, objective performance measures of the proposed method are compared with those of conventional methods such as an MVDR (Minimum Variance Distortionless Response) beamformer and an ICA (Independent Component Analysis) method. As a result, it is shown that the proposed separation method has better separation performance than the conventional separation methods.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.42 no.4 s.304
• /
• pp.111-118
• /
• 2005

In this paper, we employ the modulation technique for improving the characteristics of beamformer in the low frequencies and thus improving the overall noise reduction performance. In the 1-dimensional uniform linear microphone arrays, we can suppress the narrowband noise component using the delay-and-sum beamforming. But, for the wideband noise signal, the delay-and-sum beamformer does not work well for the reduction of low frequency component because the inter-element spacing is usually set to avoid spatial aliasing at high frequencies. Hence, the beamwidth is not uniform with respect to each frequency and it is usually wider at the low frequencies. In order to obtain the beamwidth independent of frequencies, subarray systems[1][2][3][4] and multi-beamforming[5] have been proposed. However these algorithms need large space and more microphones since they are based on the theory that the size of the array is proportional to the wavelength of the input signal. In the proposed beamformer, we reduce the low frequency noise by using modulation technique that does not need additional sensors or non-uniform spacing. More Precisely, the array signals are split into subbands, and the low frequency components are shifted to high frequencies by modulation and reduced by the delay-and-sum beamforming techniques with small size microphone array. Experimental results show that the proposed technique Provides better performance than the conventional ones, especially in the low frequency band.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2017.11a
• /
• pp.44-45
• /
• 2017

본 논문에서는 비균등 선형 마이크로폰 어레이를 활용한 convolutional neural network (CNN) 기반의 음원분리 방법을 제안한다. 우선, 주어진 어레이 배치에 따라 채널간의 시간차를 분석하고, 분석된 시간차에 따라 주파수별로 방사각과 넓이에 따라 입력 오디오 신호의 spectral magnitude를 예측한다. 그러고 나서, CNN 분류기로부터 최적의 방사각과 넓이를 선별하고 이를 통해 음원을 분리한다.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2015.11a
• /
• pp.21-22
• /
• 2015

현재 UHDTV (Ultra-High-Definition TV) 시대에 사용될 멀티미디어 부호화로 MPEG-H를 표준화로 진행하고 있다. 향후 방송용 오디오 콘텐츠는 채널 오디오 콘텐츠에서 진화하여 객체 오디오 콘텐츠까지도 필요하게 된다. 이에 따라, 본 논문에서는 고품질의 방송용 오디오 콘텐츠를 제작하기 위한 비균등 선형 마이크로폰 어레이 기반의 음원분리 방법을 제안한다. 제안된 방법은 주어진 어레이 배치에 따라 채널간의 시간차를 분석하고, 이에 따른 객체 오디오 생성을 위한 음원분리 기술을 적용한다. 제안된 기법의 성능을 검증하기 위하여 음원분리도를 측정하였고, MVDR (Minimum Variance Distortionless Response) 빔형성기와 성능을 비교하였다. 비교 결과, 제안된 기법이 MVDR 빔형성기에 비하여 12.8% 높은 음원분리도 수치를 나타낸 것을 확인하였다.