Audio-Based Musical Version Identification: Elements and challenges

In this article, we aim to provide a review of the key ideas and approaches proposed in 20 years scientific literature around musical version identification (VI) research connect them current practice. For more than decade, VI systems suffered from accuracy-scalability trade-off, with attempts increase accuracy that typically resulted cumbersome, non-scalable systems. Recent years, however, have witnessed rise deep learning-based take step toward bridging gap, yielding can realistically be deployed industrial applications. Although trend positively influences number researchers institutions working on VI, it may also result obscuring before learning era. To appreciate two decades novel facilitate building better systems, now some successful concepts applications study their evolution throughout years.