Deep learning-based musical version identification: Connecting domain knowledge with data-driven systems
By F. Yesiler

Furkan Yesiler explains to ADASP his work on deep learning-based musical version identification.

Abstract

The version identification (VI) task concerns detecting and retrieving a set of songs that originate from the same underlying musical composition. For more than a decade, VI systems suffered from the accuracy-scalability trade-off, but recent years have witnessed an increase in deep learning-based VI approaches that take a step toward bridging the accuracy-scalability gap.

In this talk, we will go through deep learning-based VI, starting with a musically-motivated method called MOVE. I will point out the design decisions and training strategies that facilitate learning semantically-meaningful embeddings for musical versions. Then, I will introduce two recent works that further improve upon MOVE regarding two perspectives: (1) the accuracy, by addressing the limitations of using only chroma features and analyzing the potential accuracy gains resulting from combining various input representations, and (2) the scalability, by addressing the large embedding sizes and proposing methods to obtain smaller embeddings while maintaining similar accuracies. I will conclude the talk by giving some ideas for the future of VI.

Bio

Furkan Yesiler is an Early Stage Researcher / Ph.D. candidate as a part of the MIP-Frontiers project (MSCA Grant No:765068) at Music Technology Group, Universitat Pompeu Fabra (Barcelona). His Ph.D. research is focused on incorporating deep learning techniques to build accurate and scalable music version identification systems for industrial use-cases. He received his MSc degree in Sound and Music Computing also at MTG, UPF with his thesis on singing voice research. He graduated summa cum laude with two BSc degrees in computer engineering and industrial engineering from Koc University (Istanbul) where he was accepted with a full scholarship. During his bachelor’s studies, he did internships in management consulting, and mergers and acquisitions advisory companies in Istanbul.