In the second article of the Metadata Matters series from Byta & MusicBrainz, we’ll dive deeper into why metadata is so critically important in a digital music world.
Miss part 1? Click here to read The Basics of Metadata.
How is metadata used to identify and credit artists, songwriters and contributors?
First, let’s examine how metadata can identify and credit artists, songwriters and other contributors. But what exactly is the definition of an artist? Artists range from one-person productions, like singer-songwriters or electronic music producers, to massive orchestras like the Philharmonia Orchestra with probably thousands of members over its lifetime, of which we have already documented 241! Or the Japanese band AKB48 that has had 459 people associated with it!
Artists can take on performing names and change those names many times in their career. For instance, Calvin Broadus has been known as Snoop Dogg, Snoop Doggy Dog and Snoop Lion. Or the artists Cassetteboy, who are known by many aliases – some apply to only a single recording!
Given this creative insanity, you can see how identifying and paying artists might become a rather tricky proposition. Before the Internet and sites like MusicBrainz, if you wanted to license a piece of music for your own project, the only way to find an artist would have been to hire an expensive entertainment lawyer. What if the people who want to use your music in their project have no money to spend on lawyers? If you can’t be found, you won’t earn money from your music.
Music sites like MusicBrainz and Discogs can go a long way towards identifying artists. MusicBrainz attempts to collect artist websites and social media feeds for artists, which all become avenues for contacting the artist themselves. However, the data in MusicBrainz, while rather extensive, may not be complete or fully up to date. In the end these sites should be used to generally inform people about music and where to find artists, but it is imperative to contact the labels/artist themselves and verify the information present online.
The importance of accurate and complete metadata
MusicBrainz tracks 30 artists with the name John Williams. Most people will think about John Williams the famous Star Wars composer (among his many other works), or John Williams the famous classical guitarist from Australia. But what about John Williams the little known English fiddler and conductor? What if you’re tasked with paying the latter John Williams, but don’t know how to contact him? All too often the industry doesn’t try very hard and just ends up paying the most famous John Williams. How often does money owed to the 28 John Williams end up in the pocket of one of the two famous John Williams? All too often, sadly.
This is why the world uses Unique Identifiers – these cold, hard numbers that feel inhumane to apply to a creative person – e0a3255c-5624-417d-bddb-82d1aa59b296 is not a sexy way of saying, “John Williams, the fiddler/conductor”, is it? But what is sexy is that it unambiguously identifies one single artist who owns rights or should get paid. That is sexy to us data nerds!
If the entire music industry could use one set of unique identifiers across the board we could ensure that more artists get paid correctly. This would allow streaming companies to accurately pay artists for every single stream, something that doesn’t always happen today.
If correct metadata doesn’t exist, then people must play guessing games about who created the music and who should get paid. At times when releasing new music, permission from other artists, for important things like clearing samples (read: acquire a license to a sample of someone else’s recording) is required. What if that artist cannot be found? This can create bottlenecks in getting music released, because releasing without the required licenses is surely going to land someone in court!
The use of metadata in music recommendation algorithms
The modern music industry mostly plays out online – the record shops where we used to find new music have given way to the many bewildering online offerings of today. This means we no longer have those die hard music fans lurking in record shops to whom we might ask questions about the latest releases.
Today music is served up by recommendation engines which are driven by opaque algorithms that mere mortals are not allowed to look at, let alone be expected to understand. These cold, inhumane systems attempt the very human task of selecting and arranging recordings with the hope that they will please human listeners.
The tricky part of this is the minimal metadata we described in article 1 does not provide the recommendation engines any sort of context about the music. Without knowing the genre of the music, when it was released, its tempo, its key, the recommendation algorithms have little data to go on. Some systems can analyse the music and determine these things from the audio itself, if the audio is available. But even these systems may not work well because music can vary so widely.
What is the BPM of an ambient electronic track? Or of Beethoven’s 5th? These concepts don’t readily apply to all types of music. Machine learning systems can quickly make an unintelligible mess that can lead to poor recommendations.
Because of this, artists should attempt to provide more context around their music: What genre is this music? When was it released? Does it have a key, BPM? If so, what are those? Does it have vocals and in what language? All of these pieces of data provide recommendation engines more context and allow your music to be recommended to others. And once people start playing your music, you should start getting paid for that!
The use of metadata in music production and recording processes
The most critical time for collecting music metadata happens during the creation process of music. Each piece of music should ideally have documentation that lists and identifies the people who have contributed to the creation of the music and their fair share of the proceeds (the splits).
This sounds fairly simple and straightforward, but the reality of this is quite messy. Clearly it is important to track the artists who performed a track live or on a recording. But anyone involved in the process should also be noted – such as all the engineers (sound, recording, etc). If anyone is earning money as part of the creation process, they should be noted – including photographers, hairdressers, travel agents and even drivers.
While this may seem overkill, one should note that having well documented records is key to paying people fairly and avoiding costly court cases where someone involved in the creation of a piece of music wasn’t adequately compensated for their work. But even historical perspectives can become quite interesting many years later. For instance, I would love to watch an interview with an errand person who was working at the studio when Pink Floyd recorded Dark Side of the Moon. The person making coffee and fetching sandwiches often has insights into these historical moments that no one else has and all too frequently they are lost in time. Did Syd Barret really randomly appear during the recording of Wish You Were Here? Is that true?
Artists create culture when they create music. We can never predict if any given recording will be forgotten or an inspiration for future generations, but we should aim to document this creation process in hopes that we can all look back fondly 20 years later and hear the behind the scenes tales that underpin the music that moves our lives. And so that everyone who should be, is fairly compensated.
This article is the second instalment of a 3-part series, Metadata Matters, brought to you by Byta and MusicBrainz. Stay tuned for part 3. Coming soon!