Abstract
The extraction of the beat from musical audio signals represents a foundational task in the field of music information retrieval. While great advances in performance have been achieved due the use of deep neural networks, significant shortcomings still remain. In particular, performance is generally much lower on musical content that differs from that which is contained in existing annotated datasets used for neural network training, as well as in the presence of challenging musical conditions such as rubato. In this paper, we positioned our approach to beat tracking from a real-world perspective where an end-user targets very high accuracy on specific music pieces and for which the current state of the art is not effective. To this end, we explored the use of targeted fine-tuning of a state-of-the-art deep neural network based on a very limited temporal region of annotated beat locations. We demonstrated the success of our approach via improved performance across existing annotated datasets and a new annotation-correction approach for evaluation. Furthermore, we highlighted the ability of content-specific fine-tuning to learn both what is and what is not the beat in challenging musical conditions.
Highlights
Along-standing area of investigation in music information retrieval (MIR) is the computational rhythm analysis of musical audio signals
We focused on one such approach, transfer learning, through which knowledge gained during training in one type of problem is used to train another related task or domain [33]
With the long-term goal of integrating in situ fine-tuning within a user based workflow for a given piece of music, we considered this aspect of efficiency to be important, and this formed a secondary motivation to extend the temporal convolutional network (TCN)-based approach
Summary
Along-standing area of investigation in music information retrieval (MIR) is the computational rhythm analysis of musical audio signals Within this broad research area, which incorporates many diverse facets of musical rhythm including onset detection [1], tempo estimation [2] and rhythm quantisation [3], sits the foundational task of musical audio beat tracking. The pursuit of computational beat tracking is not limited to emulating an aspect of human music perception. Rather, it has found widespread use as an intermediate processing step within larger scale MIR problems by allowing the analysis of harmony [7] and long-term structure [8] in “musical time” thanks to beat-synchronous processing. In particular for musicological and creative applications, the need for very high accuracy is paramount as the quality of the subsequent analysis and/or creative musical result will depend strongly on the accuracy of the beat estimation
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
