If you are using the sequential labels of DCASE 2018 Task 4 1578 audio clips, please citing our paper:
Y. Hou, Q. Kong, J. Wang and S. Li “ Polyphonic audio tagging with sequentially labelled data using CRNN with learnable gated linear units”. IEEE AASP DCASE2018 Workshop, 2018.
Please feel free to contact me if you have any questions.
Mail: hyb@bupt.edu.cn