%0 Conference Paper %1 du2021bytecover %A Du, Xingjian %A Yu, Zhesong %A Zhu, Bilei %A Chen, Xiaoou %A Ma, Zejun %B Proceedings of the International Conference on Acoustics, Speech and Signal Processing %D 2021 %I IEEE %K cover deeplearning dnn mir music similarity uncovr %P 551--555 %R 10.1109/ICASSP39728.2021.9414128 %T Bytecover: Cover Song Identification Via Multi-Loss Training %U https://ieeexplore.ieee.org/abstract/document/9414128 %X We present in this paper ByteCover, which is a new feature learning method for cover song identification (CSI). Byte-Cover is built based on the classical ResNet model, and two major improvements are designed to further enhance the capability of the model for CSI. In the first improvement, we introduce the integration of instance normalization (IN) and batch normalization (BN) to build IBN blocks, which are major components of our ResNet-IBN model. With the help of the IBN blocks, our CSI model can learn features that are invariant to the changes of musical attributes such as key, tempo, timbre and genre, while preserving the version information. In the second improvement, we employ the BN-Neck method to allow a multi-loss training and encourage our method to jointly optimize a classification loss and a triplet loss, and by this means, the inter-class discrimination and intra-class compactness of cover songs, can be ensured at the same time. A set of experiments demonstrated the effectiveness and efficiency of ByteCover on multiple datasets, and in the Da-TACOS dataset, ByteCover outperformed the best competitive system by 18.0%.

@inproceedings{du2021bytecover, abstract = {We present in this paper ByteCover, which is a new feature learning method for cover song identification (CSI). Byte-Cover is built based on the classical ResNet model, and two major improvements are designed to further enhance the capability of the model for CSI. In the first improvement, we introduce the integration of instance normalization (IN) and batch normalization (BN) to build IBN blocks, which are major components of our ResNet-IBN model. With the help of the IBN blocks, our CSI model can learn features that are invariant to the changes of musical attributes such as key, tempo, timbre and genre, while preserving the version information. In the second improvement, we employ the BN-Neck method to allow a multi-loss training and encourage our method to jointly optimize a classification loss and a triplet loss, and by this means, the inter-class discrimination and intra-class compactness of cover songs, can be ensured at the same time. A set of experiments demonstrated the effectiveness and efficiency of ByteCover on multiple datasets, and in the Da-TACOS dataset, ByteCover outperformed the best competitive system by 18.0%.}, added-at = {2021-10-15T08:14:19.000+0200}, author = {Du, Xingjian and Yu, Zhesong and Zhu, Bilei and Chen, Xiaoou and Ma, Zejun}, biburl = {https://www.bibsonomy.org/bibtex/28fb5ec1ee89e23a6793f701fce8202df/jaeschke}, booktitle = {Proceedings of the International Conference on Acoustics, Speech and Signal Processing}, description = {Bytecover: Cover Song Identification Via Multi-Loss Training | IEEE Conference Publication | IEEE Xplore}, doi = {10.1109/ICASSP39728.2021.9414128}, interhash = {ae0d09961075fcd45a66d9f2cc0b5cc2}, intrahash = {8fb5ec1ee89e23a6793f701fce8202df}, issn = {2379-190X}, keywords = {cover deeplearning dnn mir music similarity uncovr}, month = jun, pages = {551--555}, publisher = {IEEE}, series = {ICASSP'21}, timestamp = {2021-10-15T08:15:04.000+0200}, title = {Bytecover: Cover Song Identification Via Multi-Loss Training}, url = {https://ieeexplore.ieee.org/abstract/document/9414128}, year = 2021 }