Chiu, Ching YuChing YuChiuSu, Alvin W.Y.Alvin W.Y.SuYI-HSUAN YANG2023-10-062023-10-062021-01-0110709908https://scholars.lib.ntu.edu.tw/handle/123456789/635987This letter presents a novel system architecture that integrates blind source separation with joint beat and downbeat tracking in musical audio signals. The source separation module segregates the percussive and non-percussive components of the input signal, over which beat and downbeat tracking are performed separately and then the results are aggregated with a learnable fusion mechanism. This way, the system can adaptively determine how much the tracking result for an input signal should depend on the inputs percussive or non-percussive components. Evaluation on four testing sets that feature different levels of presence of drum sounds shows that the new architecture consistently outperforms the widely-adopted baseline architecture that does not employ source separation.Beat/downbeat tracking | Computer architecture | Deep learning | Hidden Markov models | Music | Source separation | source separation | Testing | Training[SDGs]SDG11journal article10.1109/LSP.2021.30845042-s2.0-85107205606https://api.elsevier.com/content/abstract/scopus_id/85107205606