AI Is Helping Spotify Lead the Music Streaming World
Sat, April 10, 2021

AI Is Helping Spotify Lead the Music Streaming World

Spotify has become the undisputed leader in the music streaming industry. This on-demand music streaming service has been providing access to users across the globe to millions of songs and podcasts from artists since 2006 / Photo by: prykhodov via 123RF

 

Spotify has become the undisputed leader in the music streaming industry. This on-demand music streaming service has been providing access to users across the globe to millions of songs and podcasts from artists since 2006. Business of Apps, the leading media and information brand for the app industry, reported that Spotify was founded in response to the growing challenge of online music piracy in the early 2000s.

As of 2019, Spotify has more than 217 million monthly active users worldwide. Of these, 100 million are Spotify Premium subscribers. The music streaming platform has had a consistent growth rate for the past years, with a 44 million increase in users (25 percent) and a 25 million increase in subscribers (32 percent) annually. It was expected that in the second quarter of 2019, Spotify’s total monthly users will rise to 222 million to 228 million. The company is also hoping to report 245 million to 256 million monthly active users by the end of the year.

Spotify is a major proof that the music-streaming market is booming. Finances Online, the fastest-growing independent software review platform, reported that music streaming services are gaining more popularity for several reasons, as cited by survey respondents: the variety of music available on them (81 percent), low price point (80 percent), ability to listen on multiple devices (68 percent), clean interface (66 percent), and good algorithm to find music (58 percent).

Spotify is leading the music-streaming industry by utilizing artificial intelligence, algorithms, machine learning, and other technology to use its subscribers’ music and personal details to shape their entire listening experience. Indeed, if there’s one thing Spotify excels at, it is knowing their consumers. As a result, users have chosen them over other giant music-streaming platforms.

BaRT for Spotify Home Screen

Spotify is a data-driven company. This makes them different from its competitors like Google Music, Amazon Prime Music, and Apple Music. Spotify has a unique level of customization and expansion of music knowledge offered to customers while the other companies rely on a mix of paid humans and community-created playlists. 

The music streaming platform has been using algorithms to drive the user’s listening experience. A great proof of this is the home screen of the Spotify app. Mounia Lalmas-Roelleke, Spotify Research director, stated that the home screen is being governed by an AI system called “Bandits for Recommendations as Treatments” or BaRT. The system, which aims to quickly help users find the music or podcast they are going to enjoy listening to, organizes each user’s home screen in a personalized way.

BaRT is the reason why the Spotify app is arranged in rows of playlists, following the theme like “keep the vibe going” or “best artists.” According to OneZero, a new Medium publication about tech and science, the algorithm tracks music to present based on the user’s recently played songs. Usually, this will only take more than 30 seconds to do it—Spotify’s key signal for understanding whether you like a song or not. But the longer you listen to the recommended playlist or set of songs, the better the recommendation is determined to be. Matthew Ogle, Spotify product director, stated that skipping before the 30-second mark is the equivalent of a thumbs-down for the Discover Weekly playlist.

Spotify is a data-driven company. It has a unique level of customization and expansion of music knowledge offered to customers while the other companies rely on a mix of paid humans and community-created playlists / Photo by: Tommaso Altamura via 123RF

 

Using Machine Learning

Another great feature that Spotify is famous for is its ability to provide personalized recommendations and help users discover new music, which is powered by machine learning. This comes as no surprise since the company has been committed to continue to invest in artificial intelligence and machine learning capabilities to deepen the personalized experience that they offer to their users and that “this personalized experience is a key competitive advantage.”

According to HBS Digital Initiative, the hub for tech at Harvard Business School building community and expertise around digital transformation, Spotify employs three types of machine learning working to enhance its recommendation engine: collaborative filtering, natural language processing (NLP), and raw audio models. The company can provide recommendations to users based on the preferences of other users with similar tastes using collaborative filtering.

Through NLP, Spotify generates “tags” associated with each song and compares those tags with those of other songs by looking into articles, blogs, and song metadata. This type of machine learning also refines the pool of song recommendations by analyzing which artists or songs are frequently mentioned along with the song in question. Lastly, Spotify can identify commonalities between songs through its musical elements through raw audio processing. 

Another great feature that Spotify is famous for is its ability to provide personalized recommendations and help users discover new music, which is powered by machine learning / Photo by: everythingpossible via 123RF

 

Predicting What Songs Will Gain Popularity

A recent study was conducted by two students and researchers at the University of San Francisco (USF) aimed to see if hit songs shared similar features. At the same time, they wanted to know whether those features could be used to predict which songs would be hits in the future. The researchers used the Spotify Web API to collect data for 1.8 million songs and collected approximately 30 years' worth of data from the Billboard Hot 100 chart.

After that, the researchers trained and evaluated four different models—a logistic regression, a neural network, a support vector machine (SVM), and a random forest (RF) architecture—to analyze a variety of song features. According to Tech Xplore, an online site that covers the latest engineering, electronics, and technology advances, they also conducted a series of evaluations to test how well the four models could predict billboard hits. 

The researchers discovered that the highest precision rate (99.53 percent) was attained by the SVM architecture. The study concluded that all four models can be used to predict a billboard hit based on features of a song's audio, including tempo, key, valence, etc. These algorithms can help record labels to determine the factors that might contribute to a song’s success, and at the same time, for users to know the type of sound they seek, reducing the number of songs they have to consider.

Indeed, Spotify continues to gain success through AI algorithms and machine learning. This opens an opportunity not only to other music streaming platforms but also other companies to provide a better user experience for their market.