Editor’s Summary

While digital music distribution has been available since the early 2000s, online platforms have been unable to support classical music. In contrast to the relatively uncomplicated metadata for popular music, the complexity of required metadata for classical music, double that for other genres, creates demands that distributors have been unable to meet. Manual metadata entry by trained musicologists is extremely costly and delays availability. DART is a music distribution platform that automatically generates metadata for classical music based on guided data input by the user combined with platform-generated data. Developed by a multifaceted musician with strong technology experience, the solution automatically delivers metadata for classical music to enable proper organization and accurate search and retrieval. The analysis of the structure of classical music that underlies DART has been adopted as an industry standard.

Keywords

music
metadata
automatic classification
database producers


Case Study:  How a Startup Solved a 10-Year-Old Music Metadata Problem

by Richard Jacobson

DART, a new music tech startup, has finally cracked the code on a long-standing problem with classical music metadata. This discovery led to an invitation to Project Music, the nation’s first music tech accelerator, and eventually opened the door for DART to help the music industry as a whole. In this article, I will tell you how we solved this metadata problem and why the solution will impact an entire genre of music.

The Music Metadata Monster

With the advent of the iTunes Music Store (now the iTunes Store) in 2003, digital music distribution was born. Shortly thereafter, online distributors like TuneCore emerged, allowing almost any artist, with or without a label, to submit music to online stores. Such technologies democratized distribution and opened the floodgates for independent artists in nearly every music genre except one. In spite of all these amazing innovations that began over a decade ago, classical artists are still struggling to get their music into online stores and streaming services.

That’s not to say you can’t find plenty of classical music in iTunes, Google Play, Amazon, Spotify and more. However classical artists who want to distribute music through online services still face technical and financial hurdles that artists in other genres do not. Today, some classical artists easily pay 10 to 100 times as much for music distribution as artists in any other genre. Why? Because the metadata for classical music is much more complex than the metadata for other styles of music. So what exactly is the metadata for classical music?

Why Is Metadata a Problem for Classical?

Put simply, metadata is data that describes other data. When it comes to digital music, it’s the metadata that tells your iPhone or iPod the name of the artist, album and song title for each music file. Without correct metadata, proper organization, accurate search and many other processes aren’t even possible. However, classical music metadata requires more information than merely artist, album and song title. Classical music often needs to contain information for a specific composer, conductor, orchestra, soloist, ensemble and more.

Yet it’s not just the sheer amount of data that’s the biggest problem for classical music. It’s the formatting that’s especially challenging. Please keep in mind Beethoven was writing symphonies long before the modern music industry gave us the single. So whereas the title for a current day single can be something quite simple, such as Taylor Swift’s Shake It Off, the title for a major classical work might be something more like Beethoven: Symphony No. 3 in E-Flat Major, Op.55 “Eroica.”

FIGURE 1.

FIGURE 1.

One reason a classical music title is often much longer than a country music title is because a classical title often contains important organizational information necessary for discoverability (Figure 1). Additionally the information must be ordered in a very specific sequence. For the above example, the sequence is composer, type of work, number, key, catalogue reference number and finally the name of the work if there is one. Furthermore, for online stores and streaming services to correctly recognize each segment of a classical title sequence, the punctuation, abbreviations and capitalization must be entered precisely. A missing or misplaced colon is all it takes to cause a serious error.

FIGURE 2.

FIGURE 2.

A Very Archaic Workaround

Obviously, getting these metadata details exactly right requires very specialized knowledge, so classical music distributors have historically employed teams of trained musicologists to manually enter the metadata needed to distribute classical music correctly. As you can imagine, the process of manually entering the data for every single classical track is time-consuming and labor-intensive, and classical artists often pay an outrageous 35-50% of their royalties to cover distribution costs. Additionally, even with trained musicologists at work, manual data entry is quite error-prone, which is why online distribution is sometimes delayed by several weeks, missing an artist’s release date altogether.

FIGURE 3.

FIGURE 3.

So why not simply use an automated distribution platform like CD Baby, TuneCore or DistroKid? Because of the sheer complexity of the metadata, none of these services will even accept a classical music submission. And that’s why we built the DART platform. It is the first fully automated distribution platform that correctly generates classical music metadata automatically.

Half Artist, Half Technologist

DART is the brainchild of Chris McMurtry, a composer who founded a classical label after working eight years for Apple. He is not only the CEO of DART, he is basically the embodiment of the company, someone with equal passions for both classical music and technology. Moreover, he is also a guitarist for an independent rock band. That background means that, before DART, Chris had plenty of experience using various automated distribution platforms that serve other genres. Whenever he wanted to submit an original piece of rock music to online stores and streaming services, he had multiple options. But whenever he wanted to submit one of his original classical compositions, he had none.

FIGURE 4.

FIGURE 4.

McMurtry found this infuriating, especially since he could clearly envision the technology necessary to solve the problem. Eventually his own personal frustration led him to create DART. However, to build a customer-facing platform, we had an even bigger challenge than simply generating correct classical metadata; we had to figure out a way to correctly generate the metadata from user input. Even though these users might be expert classical musicians, they might not be experts in musicology. So how do you build a platform that can take the information provided by a musician and use it to automatically output metadata on par with a musicologist?

The DART Platform
FIGURE 5.

FIGURE 5.

First, we examined the structure of each type of classical work (Figure 2) and then broke the structure down into a series of discrete fields for collecting user input (Figure 3). In this way, each time a user enters information into a field (Figure 4), the DART platform knows where that piece of information belongs in the sequence (Figure 5). Furthermore the DART platform is dynamic. It asks follow-up questions based on user input until it has a clear understanding of the input. It then runs the user-generated information through a complex series of algorithms to determine what is missing, then fills in the gaps (Figure 6). By combining user-generated data with platform-generated data, we get perfectly generated metadata every single time.

FIGURE 6.

FIGURE 6.

Our service represents the first time in the 10-plus years of digital distribution that someone has finally cracked the code on how to automate the generation of classical music metadata. This means classical artists who have been overpaying for distribution now have a more affordable

solution, and classical artists who previously had no way to distribute their music can finally enter the marketplace. Solving this problem

eventually led to an invitation for DART to write the style guide on classical music metadata for the Music Business Association. And we are very excited that this effort will lead to better classical music metadata across the whole industry, not just for DART customers.


Richard Jacobson is chief marketing officer for DART, one of the first eight startups selected to

participate in Project Music, the country’s first music tech accelerator in Nashville, Tennessee. Not only is DART’s CEO Chris half artist, half technologist but so is the whole team; our CEO, CTO, CMO and COO are a composer, rocker, artist and DJ, respectively. If you’d like to learn more about us and our work, please reach out to richard@dartmusic.com or check us out at dartmusic.com.