Taste Profile Attributes Go Public

October 9, 2012

If you’re reading this, you’re probably somewhat familiar with our Taste Profiles service. Taste Profiles allow us to maintain a detailed understanding of someone’s music activity - not only what they’re listening to, but also their likes, dislikes, skips, and bans. We apply Taste Profiles to help streaming services, social networks, and app developers craft the best experience for each of their users.

It’s almost hard to remember now, but people’s music data used to be hard to come by. The data scarcity of just a few years ago has given way to data deluge.

Music activity tickers, scrobbling (creating lists of what people play), social music recommendations and shares, playlist creation – the sheer volume of options to sift through quickly creates headaches for any ambitious music product or service provider.

  • Out of a user’s 300 friends, whose music activity is specifically interesting to her?
  • The next song to play will be from a Billboard Hot 100 artist. Will that keep a particular user engaged, or turn him off?
  • Is the artist someone played on Tuesday morning a good choice for Saturday night?

We feel that the music app ecosystem still has a long way to go in understanding you, as a music fan, and then applying that understanding to your online listening experience.

Today, we are pulling back the curtain on the first batch in a series of Taste Profile Attributes - a set of scores and summaries that provide a better understanding of each music fan on any service. Here are the attributes we’re making available to start:

  • Diversity: Measures the overall diversity of a fan’s listening by mapping the distance across the musical styles enjoyed by the listener.
    • Some people treat music like a buffet, bouncing between samples of jazz, speed metal, and R&B. Others prefer digging into a single, huge course of alt-country. The diversity attribute identifies what kind of fan a user is.
  • Mainstreamness: Measures the overall familiarity of a user’s listening activity to determine preference for either mainstream or more obscure music.
    • A listener may take pride in finding the deepest, most obscure cuts, or prefer to keep up to date with the Top 40. The user mainstreamness score captures this distinction.
  • Freshness: Measures listening habits to determine how much a user cares about new album releases vs. sticking with older music.
  • Adventurousness: Measures a listener’s openness to music outside their comfort zone.

We use these attributes to paint a detailed picture of each user. So far, we have only used them internally. Today, The Echo Nest offers early access to our Taste Profile Attributes for our customers and app developers.

These attributes can improve the listening experience of any music fan who uses your product. To whet your appetite, some possibilities:

  • Use the ‘mainstreamness’ attribute to find fans who spend most of their time listening to deep tracks. Instead of showing devout hipsters the new Billboard hot 100 every time they visit your New Releases section, show them what’s just starting to surface on the Hype Machine or Pitchfork.
  • Use the ‘freshness’ attribute to identify those listeners most interested in new music. When their favorite artists release new music, make sure their stream includes those songs from the moment those albums drop.
  • Let ‘adventurousness’ fine-tune any playlist for any individual listener. Give the more adventurous ones variety, by mixing in unfamiliar artists and deep tracks from familiar artists. At the same time, keep non-adventurous listeners in a ‘safe-listening zone’ surrounded by music they know and love.
  • Combine ‘diversity’ and ‘adventurousness’ to find highly-adventurous music fans with the most diverse music taste and introduce them to new styles of music.
  • Combine any Taste Profile attributes to segment your listeners into categories like these:
    • the Top-40 fan (high mainstreamness, high freshness),
    • the hipster (low mainstreamness, low freshness), and
    • the musically-stagnant fan (low mainstreamness, low freshness, low diversity, and low adventurousness).
  • Use categories not only to improve features, but also to identify fans who are most likely to subscribe to a service.

The possibilities are limitless, and applicable across a broad range of music apps and services. By putting Taste Profile Attributes in the hands of developers, we hope they can spend less time calculating weights and parameters, and more time building features to create better experiences for every listener.

To make Taste Profiles easy to integrate into existing or upcoming music services, we’re also including some frequently-requested Taste Profile summaries: Top Terms, Top Artist Years, and a summary of Acoustic Features from across the whole Taste Profile (including energy, danceability, and tempo).

If you’re interested in getting access to Taste Profile Attributes, visit our developer site.