상세정보
학술지
배경 음악 분리를 위한 확장된 합성곱을 이용한 멀티 밴드 멀티 스케일 DenseNet
- 저자
- 허운행, 김혜미, 권오욱
- 발행일
- 201911
- 출처
- 한국음향학회지, v.38 no.6, pp.697-702
- ISSN
- 2287-3775
- 출판사
- 한국음향학회
- 협약과제
- 19KS1100, 음악 및 동영상 모니터링을 위한 지능형 마이크로 식별 기술 개발, 박지현
- 초록
- We propose a multi-band multi-scale DenseNet with dilated convolution that separates background music signals from broadcast content. Dilated convolution can learn the multi-scale context information represented by spectrogram. In computer simulation experiments, the proposed architecture is shown to improve Signal to Distortion Ratio (SDR) by 0.15 dB and 0.27 dB in 0dB and ??10 dB Signal to Noise Ratio (SNR) environments, respectively.
- KSP 제안 키워드
- Background music, Computer simulation(MC and MD), Context Information, Dilated Convolution, Multi-scale, Signal noise ratio(SNR), Signal to Distortion Ratio(SDR), Signal-to-Noise, Simulation and experiment, multi-band