Wander Above the Sea of Fog

We use music to connect with others, to find solace and comfort in the absence of social support, to focus, to attain a deeper sense of meaning or perspective in life, and to transcend reality. Most of us induce these states by selecting music from a particular style, artist, or track. But today, more music is at our disposal than ever before: approximately 50k new tracks are uploaded to streaming services everyday . This vast ocean can be overwhelming: where do you even begin to look for new music?

A wise boat captain uses a map and compass to navigate the seas. Her map shows what destinations are available and her compass guides her towards the destination she wishs to visit. But can this be translated to music? Yes! Specifically, we can use data visualization techniques to discover and organize both new and familiar musical styles, artists, and tracks! Throughout this site I will be "mapping" music in an effort to make the vast body of musical work on Spotify more accesible and enjoyable to navigate.

1

A Brief Introduction to the Music Universe

This NASA Hubble Space Telescope image of the cluster Westerlund 2 and its surroundings.

One of the few constants across all cultures is music: it transcends time, location, language, wealth, social status, and more. Research suggests that people spend between 15% and 25% of their waking lives listening to music . What's more, numerous studies show that music has immediate effects on cognition, affect, and the brain

In the prelude, I mention that music streaming services upload approximately 50k new tracks every day . Naturally, we might think this would resolve the "stale playlist problem" Have you ever created a playlist you absolutely loved that eventually got boring? If so, then you've experienced the "stale playlist problem". It can occur when you listen to the same tracks you love over and over again, causing your tracks to "go stale". The stale playlist problem is common in the modern music streaming era.. But, inintuitively, this is not the case: the sheer size of music streaming databases can be overwhelming. Where do you begin to look for new artists/albums/tracks without spending a considerable amount of time sorting through the noise? Is this even possible? It probably feels like it isn't if you don't know the scope of the music space you're in. This is where "Music Atlas" can be useful: it enables you to visualize the music universe My hope is that Music Atlas enables you to discover "musical galaxies" that interest you .

What is music?

Fundementally, music is a combination of sounds, and sound is vibration Examples of sound waves seen in music. . One of the most succinct definitions of music comes from the Italian composer Ferruccio Busoni, who said, "Music is sonorous air." What's remarkable is that combinations of these simple vibrations, which are invisible to the human eye, can facilitate a deeply rich emotional experience, alter perception and consciousness, and induce ecstatic states of being .

There are countless ways to combine "sonorous air" to create a musical piece Within the scope of Western popular music, a few ambitious researcher have tried to count or create all possible melody combinations (exlcuding rhythmic differences) . If every piece were unique⁠—⁠sharing no similarities in melody, rhythm, timbre, and pitch with any other piece⁠—⁠we would have a much more difficult time discovering and organizing music. Fortunately, common features emerge: sets of tracks may share rhythmic, timbre, pitch, or lyrical themes. We can use these shared themes to create music ontologies and taxonomies Ontologies and taxonomies are in many ways similar: they describe types of things, and are arranged in a hierarchical structure. The way they differ is in their level of formality of their semantic relationships: taxonomies tend to relate classes using only is_a: kidney is_a organ is_a anatomical structure whereas ontologies use a more expressive set of pre-defined relationships such as part_of, develops_from, etc.

Music Ontologies & Taxonomies

Music is often organized in an ontological or taxonomical structure. These structures will never perfectly describe music. Just like any language is inadequate to express emotions, so are ontologies/taxonomies imperfect tools to describe music. Yet, they are still far better than nothing, and very useful when they are designed well. The most common and well-known approach is to organize music through genre ontologies/taxonomies .

The Definition of Music Genre

The traits that define a genre are more than a similar sound or summary of technical elements: subculture, fashion, geography, mentality and period of time all qualify as possible characterists of which a genre might be recognized . We can think of music genres as galaxies existing in the "musical universe". Of course, this universe doesn't really exist⁠—⁠if it did, I'd be there! But, I find this metaphor helpful to understand music ontologies/taxonomies.

Astronomers describe the Universe using a hierarchical structure with several levels: I follow a similar approach to explain the musical universe. The first significant level is the "genre galaxy," represented by a common denominator among a "large enough" What is large enough? There is no right or wrong answer: it depends on who you ask. group of music (not necessarily artists) connected by specific instrumentation, style, mentality/ideology, sound, place and/or time. Next, there is the "genre galaxy cluster," represented by a group/cluster of genre galaxies that share a common denominator. Beyond the scale of genre galaxy clusters, there is the "genre-supercluster," consisting of several galaxy clusters and groups bound to one another by, you guessed it, a common denominator.

The unresolved question is whether there are even larger structures that lie somewhere in between genre-superclusters and the musical universe as a whole. But for this work, I think it is safe to consider genre-superclusters as the second-highest level in the hierarchy.

The Hitchhiker's Guide to the Music Atlas

Music Atlas comes with three "maps"The number of visualizations may change as the project matures.:

The Genre Map provides a "bird's eye view" of the music universe. This is the broadest map and most ambitious visualization in the Music Atlas. The Playlist Relationships Map will explore relationships between popular playlists on Spotify. This map also represents the music universe at a high level, albiet to a lesser degree than the Genre Map. The Artist Sound Progression Map will visualizes how an artist's sound changes over the course of their career.

Acknowledgments

I am grateful to Chris Olah and the rest of the Distill Pub team for there insightful blog posts. Also, a big thank you to Sophie Engle for formally introducing me to data visualization.

Contributions

Work was complete by Kai unless noted otherwise.

Concepts & Writing: Kai introduced and wrote the central concept and structure of this article.

Figures & Code: Some figures were drafted in RawGraphs. For all figures, the final version was implemented in D3. All code was written by Kai.

Reviewers

Some text with links describing who reviewed the article.