Metadata problems in the music industry and how to match music metadata

For the upcoming Berlin MIR Meetup event we are happy to welcome Sertan Şentürk, who will talk about problems related to metadata in the music industry and how to match them efficiently. The event will take place at the Mindspace Co-Working space at Krausenstraße 9, 10117 Berlin Mitte.

Abstract:
In recent years, music streaming services have greatly improved the accessibility of music, and we are also observing a surge in music listening as a consequence. This trend has also been driving many technological developments in music, from recommender systems to intelligent music generation tools.

On the other hand, creator-facing side of the music industry has somewhat fallen behind the race. In particular, the metadata issues are arguably the most pressing problem across the music industry right now, especially when it comes to paying royalties to the creators.

In this talk, we will scratch the surface of some of the metadata problems in the music industry. I will discuss several reasons related to metadata quality, trust, data literacy, etc., and argue how we can improve over the current state from a data-driven perspective. I will mainly focus on the metadata matching problem, which is identifying the same entities and/or linking relevant entities across different datasets. I will present several solutions for various matching tasks with different requirements.

Bio:
Sertan Şentürk has been working on music information retrieval for 10 years, specializing in research & development. His work involves developing signal processing and machine learning-based methods to describe and discover music data. He received his PhD from Music Technology Group, Universitat Pompeu Fabra, Barcelona in 2017, and he is currently a data scientist in Kobalt Music Group, London.