Music Visualization as the Content Interface

This post was written for my client Spectralmind and appeared initially on their blog:

In the history of mobile music, content discovery has always been a challenge. Music stores represent a special kind of information overload. Exploring the depth of a super-sized content catalog, given the limited screen size of a mobile device, is a bit like doing the weekly shopping while looking through a matchbox cover.

Already in the days of the ringtone craze, music distributors thought of methods to improve the content exploration experience on small screens, hoping to create a discovery convenience that ultimately adds to the stickiness of the mobile storefront and that stimulates higher consumption. Since then, a flurry of content discovery approaches have been put in place.

music discovery on a mobile phone

Here’s a quick overview of some of the main discovery methods:

In the beginning was the browsing portal. Drop-off rates of more than 50% per menu level, even in popularity-based menu structures invalidated this model. The battle for main page presence was decided shortly after in favor of access categories like “new” (aka “novelties” or “latest additions”), ”most wanted” (aka “charts”) combined with the display of lists of noteworthy, editorially selected albums or tracks.

  • Personalized storefronts. The idea is to rearrange a mobile storefront according to a user’s previous browsing history, assuming that the historic session would be indicative of the user’s preferences. Users appreciate personalization, but want popular content at other times.
  • Discovery through search. Valid idea, but only if the user knows exactly what to search for.
  • Recommendation engines. An attempt to infer a user’s music preference algorithmically from his/her past purchases, followed by suggestions of music bought by other users with similar preferences. Alternatively, recommendations are derived from human classification of content as the basis for the suggestion of matching titles. Such recommender systems have a permanent place in today’s music storefronts.
  • Social sharing and communities. This idea picks up the concept of “following” (another user and his/her purchases or music plays) or the sharing of playlists and their proliferation through social networks.

What we find in today’s music stores is a best-of-breed combination of all of the above. These are tried and tested methods. Still, we believe music discovery needs an innovative push. Music services are increasingly similar. They offer more or less the same content at the same price. They even look similar in terms of their user interfaces and they provide comparable user experiences. In short: music distribution needs differentiators to avoid commoditization.

Here at Spectralmind, we believe in data visualization as the new frontier of music discovery. Data visualization is an art, which attempts to turn even very big data sets into visual patterns, structures and elements, in order to make the data readable and understandable. There is no doubt that music represents an enormous body of data. The leading digital music distributors pride themselves on managing catalog sizes in the range of 15-20 million tracks. Visualization methods can repackage such volumes into easily accessible formats.

Our approach goes beyond the static visualization of data. In sonarflow, our visual music browser, graphical catalog visualization is the interface to navigate, operate and explore vast arrays of musical content and to expose music recommendations in a spacial and gestural environment.

This interface is capable of embracing core user needs for content discovery:

  • browsing through large stocks of content in an intuitive and seamless way
  • discovery through serendipitous expedition, ready to encounter music of unexpected relevance
  • personalization through playlist creation
  • social sharing

So hey, if you are in music distribution, don’t fall into the commodity trap. Get in touch, we would love to show you our approach.

Search Ain’t Misbehavin’

This post was written for my client Spectralmind and appeared initially on their blog:

Searching music offside of the mainstream can be tedious. Recently i fell for a particular jazz piano genre, called “Harlem Stride Piano” while listening to a radio broadcast. Stride piano developed in the 1920s and 1930s in New York as an advancement from Ragtime. It is characterized by a rhythmic left hand play, where the pianist alternates a bass note or octave on the first and third beat with chords on the second and fourth beat, while the right hand plays the melody line. This causes the left hand to leap great distances on the keyboard, often at neck-break speed. Back then, pianists like Fats Waller, James P. Johnson or Eubie Blake were famous stride virtuosos.

Louis Mazetier introduces harlem stride piano

Today, only a few pianists are capable to play stride, and I was curious to find out about contemporary ”Harlem Stride Piano” interpreters and recordings.

The textual search for “Harlem Stride Piano” in iTunes led to zero results. Even in the advanced search of iTunes, you can only search for artists and interpreters, title- or track names, but not for genres. A search just for “stride piano” brought up one album, fortunately carrying both terms in its title. Similar, Spotify´s search for “Harlem Stride Piano” did not match anything, whereas a search for “stride piano” returned a few albums because of the use of the terms “piano” and “stride” in their titles or tracks.

Still unsatisfied, i continued the search for contemporary stride players in Google, YouTube and Wikipedia to find out about artists like Louis Mazetier, Günther Straub or Bernd Lhotzky. Knowing their names finally helped me to find the desired tunes in iTunes and Spotify.

This little research clearly depicts the limits of text based music search. It´s results depend largely on the coincidental presence of the chosen search terms in the title or artist name. If you have nothing but a tune, search is often impossible. What´s missing is search for music based on the sounds of a sample track.

While chasing contemporary “Harlem Stride Piano” records through Spectralmind´s audio intelligence platform, I certainly would have used Fats Wallers “Ain´t Misbehavin“. For sure, a sound-similarity search would have brought up more and better results in far less time.

Musing About Music Similarity

This post was written for my client Spectralmind and appeared initially on their blog:

When we demo Spectralmind’s SEARCH by Sound, a similarity search engine for music, we often realize how different the focus is on certain aspects of “similarity” among listeners. The similarity results calculated by the Spectralmind platform appear “similar” to one listener, but are judged as “not similar” by another or “somewhat similar” by a third.

Musical similarity is a very complex area and the reason for the deviations in judgement stems from the fact that similarity has so many dimensions. This raises the question, to which dimension do people relate when asked about the similarity of music?

Personally I observe that people try to exemplify similarity first of all from melody. The particular succession of higher and lower tones that form a melody is clearly a distinctive feature, which allows the listener to determine the degree of likeness or even closeness between two musical works.

Trombone Shorty at the Jazzfest Wien, 2011
Trombone Shorty at the Jazzfest Wien, 2011

But there are other dimensions of similarity as well:

  • Timbral similarity: timbre refers to the the tone color of a sound, which varies significantly among the characteristics of the sound-creating device, such as voice, string or wind instruments. As a listener we are able to identify the kinds of instruments playing, even in an ensemble like a band or an orchestra. The same melody played by a piano or a saxophone or a guitar makes a big difference in terms of timbral similarity.
  • Rhythmic similarity: rhythm is made up of a repeating pattern of sounds and silences. We perceive rhythm as fast or slow. Through rhythmic beats alone, we can set apart musical genres from each other, like rock from reggae. Music, dance and even spoken language rely on rhythm as a main and defining element. Different rhythms can be put underneath the same melody (which can be highly entertaining or massively disturbing). This practical example of melodic similarity combined with rhythmic dissimilarity highlights the difficulty to assess an overall measure of similarity between two pieces of music.
  • Structural similarity: this refers to the occurrence of specific sections within a piece of music. Common sections are intro, verse, chorus (also known as refrain), interlude and outro among many more. These are formal criteria, which can be applied to describe constructive or sequential similarities of e.g. pop music songs or symphonic compositions.

There are many more dimensions of similarity beyond the ones mentioned. Some of them are even inaccessible to human perception, but very perceptible to musical data-mining programs such as the Spectralmind Audio Intelligence Platform.
Similarity decisions need to be judged by the rationale of the similarity search. Sometimes, melodic resemblance is the searched-for attribute. In other cases it might be rhythmic conformity or timbral affinity. Or a mix of multiple qualities. The crucial factor is the intended use of the similar-sounding music. Having this intention in mind helps to escape a possible bias.

We are striving to improve our software in a way that makes its similarity opinion more comprehensible and transparent. Users have a desire to understand which dimensions of similarity the software uses to suggest something as similar.