Welcome to OpenForum. Join our global community: Sign up, Login or Learn more

The Echo Nest: the data behind personalized playlists

Millions of songs. A chance to know the unique musical and cultural aspects of each one...including which ones you will like.

Photo of Tigger
1 0

Written by

"The song-picking puppet master pulling the playlist strings behind iHeartRadio, Spotify and Nokia's music services."– Engadget

The Echo Nest can help suggest music to individuals based on listening patterns and an enormous database of digital music it can sort and recognize via audio fingerprints. To find music today, we cannot browse a physical Tower Records or Virgin Megastores and small stores are curated with a small set of music tastes in mind. When searching Spotify or iTunes, the enormous choice and lack of direction is often overwhelming. Here is where The Echo Nest enters with an enormous audio fingerprint database and an understanding of what you might want to hear.

A lovechild of science and music born out of the MIT Media Lab, The Echo Nest developed its database relying heavily on web crawling for music and for text information about songs found in blogs, websites, Tweets, etc. With this information, Echo Nest applies “natural language processing and machine learning to contextualize1” the internet’s reaction to and views on the music Echo Nest has technically analyzed separately.

The Echo Nest creates value by continuously growing its database of digital music “knowledge,” synthesizing listening patterns of users, and matching those patterns to additional music that could be appealing to the user. Consumers only experience this via digital music distribution businesses that license The Echo Nest solutions or partner in other ways with The Echo Nest, so it is not a service we encounter directly. Additionally, The Echo Nest makes available its database and an API for app developers to use with other music applications and for academic research. Consumers directly experience value created through app development that The Echo Nest supports; and consumers should benefit longer term from academic research – whether it results in more refined understanding of music taste or discovery about music therapies.

The Echo Nest offers a range of ‘solutions’ to couple its database with predictive recommendations or playlist creation. The Echo Nest’s exact value capture model remains opaque, but it appears they partner on a licensing basis for a variety of its capabilities. I imagine partnerships range from commercially using its database with the partner deciding exactly how it use the data to a fully baked partnership with Echo Nest providing a full feature or capability to a partner. Spotify acquired The Echo Nest in early 2014, which is certainly a form of capturing value. Based on its web site, The Echo Nest still partners with a large range of digital music providers, so it appears it is still capturing value outside of its subsidiary data. It will be interesting to see where this technology leads, how it might come to scare us in understanding our aural experience better than we do ourselves, and if and how Echo Nest will continue to partner outside of Spotify.

1 http://blog.echonest.com/post/52159005051/how-we-resolve-artists-on-the-internet

1 comment

Join the conversation:

Photo of Emily

I'd be curious to know how they do their data-crawl & contextualization when targeting a specific genre of music. For example, how do they filter out negative comments that are a result of a user being negative on an entire genre which would not be relevant to others who listen to that genre, versus giving weight to negative comments about a specific song within a genre? Ditto weighting of super vocal users who comment the same opinion on various blogs versus one user who comments only once. Are there key opinion leaders that the algorithm searches for and takes into account as influencers? It's such a difficult and interesting problem that they are tackling. Successful development of such web crawling and natural language processing of the text it finds has application that go way beyond music to other areas; synthesizing reviews and buying patterns in a way that is meaningful could have significant influence on product design for any type of product.