Literature Review:
1.1. Defined
Pandora Media Inc. (also known as Pandora Internet Radio or simply Pandora) is a music streaming and automated music recommendation internet radio service powered by the Music Genome Project. As of August 1, 2017, the service, operated by Pandora Media, Inc., is available only in the United States. The service plays songs that have similar musical traits. The user then provides positive or negative feedback (as thumbs up or thumbs down) for songs chosen by the service, and the feedback is taken into account in the subsequent selection of other songs to play. The service can be accessed either through a web browser or with its mobile app. Pandora is a freemium service; basic features are free with advertisements or limitations, while additional features, such as improved streaming quality, music downloads and offline channels are offered via paid subscriptions. On September 24, 2018, Sirius XM Holdings announced its intent to acquire Pandora for $3.5 billion.
1.2 Functionality
Streaming
A station is set by specifying an artist or song, or a combination of multiple items of any kind in a single station. Listeners can tune into established genre stations, other users’ stations or create their own stations based on their musical interests.[21] Each track played can be responded to with favorable (thumbs up) or unfavorable (thumbs down) buttons, which determine if it should be played, and if similarly classified songs should be played in the station. A second negative response to the same artist will ban that artist from the selected station unless the user has marked the artist positively on another occasion or if that artist is listed under the station’s variety. No response is applicable to musical attributes or to albums. An unfavorable response immediately stops play of the track. Clicking the thumbs down or skipping to the next song too many times in a row will result in a short ban of skips. Pandora also utilizes short advertisements between every couple of songs.
A menu is provided with options, such as “I’m tired of this song” (allowing a user to remove a song from the station temporarily, which counts as a skip), “Why was this song selected?” (allowing users to learn more about the composition of each song), “Move song to another station,” “New station,” and “Bookmark.” A “Buy” button is located at the top of each song block. From there, listeners can click on links to purchase the song from iTunes or Amazon.
There is a setting in each member’s account allowing the user to censor songs with explicit lyrics. The censoring option applies exclusively to song versions from albums with a Parental Advisory label, as other songs with censored versions will have that version played. An example is the song Jet Airliner by the Steve Miller Band, which had one word censored for radio play. With explicit lyrics off, that version will play, because the album did not have a PA label.
While listening, users are offered the ability to buy the songs or albums at various online retailers.[21] More than 450 musical attributes are considered when selecting the next song. These 450 attributes are combined into larger groups called focus traits. There are 2,000 focus traits. Examples of these are rhythm syncopation, key tonality, vocal harmonies and displayed instrumental proficiency.
The service has two subscription plans: a free subscription supported by advertisements, and a fee-based subscription without advertisements. There are advertisements in Pandora Mobile for mobile phones and in the Pandora computer appliance. In October 2014, less than 5 percent of active listeners were paying subscribers.[22] At the time of Pandora’s IPO in 2011, Pandora had 800,000 tracks from 80,000 artists in its library and 80 million users.[23] By November 2014, Pandora had approximately doubled its library size, topping 1.5 million songs.[24] As of end of mid-year 2018, Pandora had 71.4 million active users.[25]
In September 2016, Pandora announced additional features and subscription options, including a mid-level subscription service known as Pandora Plus that offers advertisement-free streaming, offline playback support using a prediction mechanism and more skips and replays. Users of the free advertisements-included service also were provided more skips and replays in exchange for watching an advertisement. Pandora also announced the launch of an on-demand service similar to competitors such as Apple Music and Spotify.[26][27]
On March 13, 2017, Pandora launched Pandora Premium, a new service allowing users to listen to and create playlists of individual songs on demand. Pandora’s suggestions engine suggests and recommends songs and albums, as well as generate playlists based on similar songs. Pandora also emphasized a use of machine learning and manual curation, and that it had filtered “karaoke tracks, knock-off covers and pet sounds (but not Pet Sounds) that slow down other services” from its library
1.3 Technical Aspects of Pandora
What is the Music Genome Project?
Inspired by the Human Genome Project of the 1990s and early 2000s, the Music Genome Project (MGP) was conceived by Pandora founder Tim Westergren to catalog the fundamental characteristics of the vast body of recorded music. The goal of this ambitious project was to allow music lovers to discover music based on inherent musical qualities, rather than sales data or industry-backed marketing. The Music Genome Project provides a detailed analysis of millions of songs, describing features of harmony, rhythm, melody, vocals, instrumentation, lyrics, and more. This data powers our best-in-class recommendation algorithms. Thanks to the rigorous and detailed music analysis of the Music Genome Project, none of our competitors can come close to the depth of our content understanding.
When analyzing a song, we often start with the genre, choosing from one or more of the 1300+ subgenres we have developed in our comprehensive genre taxonomy. Since many songs fit into more than one main genre, or borrow influences from a variety of genres, our nuanced analysis process allows us to indicate multiple genres and influences. Training is an essential ingredient for the Music Analyst team. In addition to extensive new hire training, ongoing training is required for maintaining broad genre knowledge, and for staying on top of the latest trends in music.
The level of depth and detail that goes into tagging a single song is unmatched anywhere in the streaming music universe. Pandora kind of analysis across millions of songs, which we then leverage to further our content knowledge across tens of millions more songs that make up our full catalog. Our data science team has used this massive treasure-trove of data to create the best music recommendation systems in the world, armed with the most comprehensive and complete content understanding database available anywhere. We believe that the next song matters, and that is why we work hard to make our streaming service the best in the world. I hope this helps explain why when that next song comes on your Pandora station, the recommendation we provide is the absolute best available
MGP2 – What’s Next for the Music Genome Project
Here at Pandora, we have recently completely redesigned the way we analyze music for the Music Genome Project, with a new system we call MGP2. We’ve developed a collection of new taxonomies that we use to describe songs, and a new, text-based tagging system that allows us to annotate music much more accurately and completely. This new way of annotating has improved all of our downstream systems and led to important improvements in the data science, machine learning, and content understanding that make our recommendation systems the best in the world.
What is MGP2?
When the Music Genome Project first began over 20 years ago, music analysis was done on pencil and paper, with analysts manually ripping CDs (remember those?) into the ingestion systems.
Music Analysts analyzing music in the early days of PandoraThe Music Genome Project Interface was soon developed, consisting of a series of “genes” that needed to be scored on a ten-point scale. This system enabled us to gather a massive amount of uniform data on millions of songs. However, as the years went by and music evolved, the original fixed set of genes could not keep up. As the tech stack aged, making changes to the system proved too costly and impractical. In the rapidly changing and increasingly genre-agnostic world of modern music, we needed a more fluid and flexible way to analyze songs.
Enter MGP2. This semantic, tag-based system lets us listen to songs and then add tags from a set of taxonomies. These taxonomies include:
• Genre
• Musicology
• Instrumentation
• Vocals
• Lyrics
• Mood
• Overall
Having fluid and dynamic control over these taxonomies allows us to add or adjust things as needed. It also turns out that these tags are much easier for humans to interpret, and also easier for machine learning models to use as inputs.
Pandora wrote thousands of translations, essentially a set of rules, to convert the numeric scores of the 2.2 million songs we analyzed in our old system into the text-based tags we created for MGP2. Once our science team updated our machine-learning models to use these tags as inputs, we saw immediate improvements in the scale and accuracy of all downstream systems, including predictions, recommendations, track grouping, and more. These models allow us to leverage the information from the 2.2 million songs we have analyzed onto the rest of the tens of millions of songs in our massive complete catalog. You can read more about the groundbreaking work they have published.
https://community.pandora.com/t5/Community-Blog/What-is-the-Music-Genome-Project/ba-p/116426
The Pandora algorithm uses these “genes” to affiliate songs on users’ stations, drawing on the user’s history and the collective history of all Pandora users. Although the relations that filiate certain songs and genres are fairly clear to Pandora users, the MGP’s genomic language and taxonomy may or may not be apparent. In keeping with the highly individualized discourse, most Pandora advertising emphasizes how the user’s individual history determines the songs on their stations.:
Pandora
While there doesn’t seem to be much word on how Spotify’s biggest competitor is approaching Artificial Intelligence, Pandora is another platform using AI to suggest music to listeners. In fact Pandora started life as the Music Genome Project, which was intended as a way to categorise music. To sort all of this music, CEO Tim Westergren and his team listen to the songs and manually assign them characteristics based on musical elements within the track. This means that for any given song, a list of other songs with similar attributes can be created. The whole concept behind it being that if someone enjoys one song, then they’ll enjoy others which share its attributes. As a way of demonstrating this in action, Pandora Radio was built, which can be accessed either through a web browser or through the dedicated mobile app. The service essentially gives users personalized radio stations aiming to have a balance of familiar and new music.
Pandora Radio makes use of a machine learning algorithm to put the Music Genome Project into practice. By using the attributes assigned to the songs alongside reactions from the listener (either a thumbs up or a thumbs down), the service is able to better decide what songs and artists to play or avoid playing.
https://futuresonic.io/discussion/ai-in-music-streaming/#:~:text=While%20there%20doesn’t%20seem,Writing%20Music%20With%20AI
Genomic Searches. Perhaps the most powerful search approach, and the one that engages students in the novice-to-expert transition, is the search based upon genomic similarity. As we think toward the future when the Genome will contain a wide range of multimedia assets from different engineering disciplines, genomic search tools become a way to illustrate the connections between seemingly disparate and unrelated problems. The collection of traits describing assets in the genome spans an n−dimensional space, and this genomic search amounts to a high-order nearest-neighbor search along the lines of Pandora’s search algorithm
The output of this search is the set of multimedia assets that are genetically related; that is, they are related based upon some underlying features defined by the controlled vocabulary of the ontology. Genetic similarity is based not primarily upon a single gene or tag, as in keyword or facet searchers, but rather on a collection of genes that are, in the aggregate, expressed in each of the related multimedia files.
Keyword and facet searches expose explicit similarities between multimedia assets (for instance, both keyword and facet searches will present multimedia assets in the same application area or that have the same underlying mathematics). However, the genomic search finds genetic near-neighbors wherever it can find them–including neighbors whose relationships are not explicitly clear..
So the Music Genome is a collection of digital assets (the songs) tagged with highly granular, descriptive attributes (the genes), and organized into a searchable database. The result is Pandora–the Internet radio station that allows users to probe the Music Genome and create playlists based upon keyword searches. Pandora interrogates the Music Genome to create a playlist of songs that are genomically related – the songs are close neighbors in the genome. Songs are not arranged by genre, or by era, or by band geographic origin, or by sales, or by any other coarse metric or hierarchical relationship. Rather, songs in the playlist are presented based upon their genomic similarity, based upon shared attributes.Initial-development-of-the-engineering-genome-project-an-engineering-ontology-with-multimedia-resources-for-teaching-and-learning-engineering-mechanics.pdf nitial-development-of-the-engineering-genome-project-an-engineering-ontology-with-multimedia-resources-for-teaching-and-learning-engineering-mechanics.pdf
References:
Initial-development-of-the-engineering-genome-project-an-engineering-ontology-with-multimedia-resources-for-teaching-and-learning-engineering-mechanics.pdf nitial-development-of-the-engineering-genome-project-an-engineering-ontology-with-multimedia-resources-for-teaching-and-learning-engineering-mechanics.pdf
https://futuresonic.io/discussion/ai-in-music-streaming/#:~:text=While%20there%20doesn’t%20seem,Writing%20Music%20With%20AI
https://futuresonic.io/discussion/ai-in-music-streaming/#:~:text=While%20there%20doesn’t%20seem,Writing%20Music%20With%20AI
https://community.pandora.com/t5/Community-Blog/What-is-the-Music-Genome-Project/ba-p/116426