What is Discover Music?
Discover Music is a way to find new music makers, or rediscover music makers you'd forgotten about. We consider the app to be successful if when you search for a music maker you know/like already and browse through the results, you see a couple of interesting-looking names to click on and check out. Keep using it and support our initiative via the Pricing tab!
Success finding interesting new music makers is what we've experienced while building this software, and one of the things that motivates us to share Discover Music beyond the University of York research environment in which it was developed.
Help and community
To get help with the Discover Music app, to report a bug, or to connect with our community of users and developers, please visit the Music Enterprise Works York Discord server and send us a message. We'll be happy to hear your feedback!
Music Maker Portal
If you make music and your name does not currently appear amongst results on the Discover Music app, but you would like it to, then please tell us about yourself and your work via the portal below, and we will determine how best to include this information in our system.
Background
Discover Music was based on an idea by Gordon Rawlins, with further research and development by Tom Collins of the Music Computing and Psychology Lab in the Department of Music, and Jude Brereton in the Department of TFTI at University of York, UK.
Since March 2020, the Discover Music app has benefited from occasional development work by Jonno Witts, Nick Moriarty, Oliver Still, and Luke George.
FAQ
Why did you develop this app? Two reasons:
- Music discovery is not a "problem" with one "solution". We don't like how commercial streaming platforms' AI music recommendation algorithms tell people what they will like. These thoughts prompt the question "Is there a way to give users more agency over discovering music makers and sharing these discoveries with one another?" Our solution still comprises AI, broadly construed, but it doesn't predict or tell a user what they will/should like;
- We think music discovery services should be informational rather than profit-driven. For a commercial streaming platform, a "successful recommendation algorithm" is one that increases revenue. Research (e.g., Hodgson, 2021) suggests this isn't in the best interests of users or music makers.
How does it work? Transparency of software is important to us, so we are happy to answer this question. When a user hits search, we run their query against several web APIs to gather information about a music maker. We analyse the results to determine music maker X is influenced by (or at least refers to) music maker Y, and/or vice versa. The results form a directed graph, which we then visualise for a user to explore, and augment with further search queries if they wish.
What do the line colours in 2D and 3D networks signify? If the line colour begins the same as the colour of the node from which it emerges, then the music maker associated with this node is influenced by or at least mentions the music maker associated with the node at the other end of the line. (We need to fix the interface so that when the influence/mention goes in both directions, the line remains the same colour.)
Is it patented? No. But there is substantial intellectual property behind this work, which is subject to copyright, and which the University reserves the right to defend if breached.
Research questions, related research and references/credits
- To what extent is it possible to give users more agency over discovering music makers and sharing these discoveries with one another?
- How can we address widespread gender imbalance in the music industry, and to what extent can music discovery services be designed to make the music industry more inclusive?
- How do informational and commercial music discovery services compare in terms of helping users to find new or local music and helping music makers to find new or local listeners?
In terms of related research and work, similar interfaces include Connected Papers and a few graph-oriented music search engines such as Musicroamer and gnod Music-map. We would also like to mention Peter Knees' nepTune project.
References
Hodgson, T. (2021). Spotify and the democratisation of music. Popular Music, 40(1), 1-17.
Acknowledgments and credits
We are grateful to York Impact Accelerator Fund and the EPSRC Impact Accelerator Accounts for funding this project.
[Some icon/image credits to be added here in due course.]