The Technology Behind The First Internet Radio
How Pandora Radio Actually Works
Back in 2000, Tim Westergren had the vision of creating a unique internet service that would enhance the way people all over the world listen to music. That’s exactly what he did by creating Pandora Radio. Along with the help of a variety of music experts, he managed to evolve traditional radio into an internet phenomenon.
What allows Pandora to continue to separate themselves from the competition is a revolutionary idea called the Music Genome Project.
Pandora’s Musical Database
The Music Genome Project, named after the Human Genome Project, is the foundation of Pandora’s product. All of the music Pandora streams to your device is a result of the work that went into the project up until the website’s creation in 2005 and still goes on today in order to establish the company’s ‘Music Genome’.
What the majority of users don’t realize however is that both the website and mobile app are the interfaces that give us access to the Music Genome Project database.
In general, the purpose of the Music Genome Project is to analyze specific characteristics of each song and find similarities between multiple songs or artists. While on the surface that sounds like a simple premise, in reality it’s a difficult process to accomplish.
As technology writer Michael Miller references in his article about Pandora, “every song in the project’s database is analyzed by a trained musicologist and assigned up to 450 unique musical attributes. These attributes are combined into larger groups called focus traits, of which there are 2,000 possible combinations, such as rhythm syncopation, key tonality, vocal harmonies, and the like.”
The Music Genome then considers these attributes and combinations of focus traits when a user creates a new station and when determining what song to play next.
Nolan Gasser, Chief Musicologist at Pandora, explains that this process allows them to take a scientific approach to dissecting music. They are able to break down each “species”, or type of music, into their basic individual components, or “genes” that make up each song’s musical identity, and then identify other songs that have a similar “genotype” or style.
Pandora even gives the user the ability to view these specific traits of every song they listen to, and discover why each song was chosen to be included on that station. An example of musical qualities the Music Genome comes up with includes “mellow rock instrumentation, folk influences, disco influences, a subtle use of vocal harmony”, along with thousands of other possible combinations.
The User Experience
If the user wants to truly be able to enjoy their listening experience though, they must design their own account according to their musical interests and tastes. In order to do this, the user must create customizable stations that identify the types of songs they want to listen to.
First, the user can simply enter the name of a specific artist or song they like in the website/app’s search bar, and click the button marked “Create a Station”. This process can be conducted a number of times and as a result, the algorithm will then begin to play music that relates to the original song or artist you started each station with.
The second way to create a station is through the use of Pandora’s new “Browse” feature.
As Lizzie Plaugic points out in her article reporting the new feature, “Browse offers selections of Pandora stations based on what users have previously listened to and the songs they’ve liked or disliked.”
“Browse” provides stations that are organized around a combinations of artists, genre, and even moods. Users can even see what songs are playing on each station to help them decide whether or not to add it to their personal list of stations.
Furthermore, there are even more steps each user can take to improve their listening experience on Pandora. One of these is the use of the “thumbs-up” and “thumbs-down” buttons. This function allows the user to physically “like” and “dislike” songs that pop up on their feed. As a result, you will begin to hear more songs like the ones previously “liked”, and less like the ones you previously “disliked”, therefore improving the ability of the Music Genome Project’s algorithm to provide you with songs that appeal more to your musical tastes.
Adding on to this, Pandora has recently added “Thumbprint Radio” to its arsenal, giving the user a more mainstream way of listening to the songs they want to listen to. Thumbprint Radio accomplishes this by taking all of the songs the user has ever “liked”, and compiling it into one playlist along with other songs the Music Genome Project database feels that you would enjoy.
As Pandora’s CPO Chris Phillips stated back in 2015, “This is a living, breathing station that will continue to change as you listen — each time you thumb up a track on Pandora, your station will update and evolve.”
Beyond Music Streaming
However, Pandora goes beyond just providing free music to its 81 million users. It has also become a place to connect consumers and businesses nationwide.
This has been accomplished through Pandora’s ad-based revenue model, where they sell advertising space to all types of business. Companies are able to successfully reach their intended target markets with over 300 audience segments on Pandora, and have the option of using either audio, banner, or video advertisements to display its products to Pandora’s users. And while the average consumer normally dislike advertisements, Pandora users generally accept this as part of the free music they are being given access to.
As you can see, Pandora can credit its success to its revolutionary Music Genome Project, easy-to-use app and website, and its ad-based revenue model. By selling advertisements, it has built connections with a variety of businesses nationwide; with its user-friendly service, it has continued to attract millions of people each year; and with the Music Genome Project it has created a truly personalized listening experience.
According to the company itself, the project is “the most sophisticated taxonomy of musical information ever collected.”