Enlarge Image

LYMO: Profiling Music And Its Emotions

When people think about data in the context of music, most likely what comes to one’s mind is how music can be recorded, published and distributed. In recent years, platforms such as Spotify and SoundCloud added the analysis of their users’ behavior, and can now give recommendations based on that.

What if there was more behind data in music?

In 2016 a task force at td sat together brainstorming about what we called Music Analytics. It soon became clear music is so much more than tunes or data. Music impacts society. Society impacts music. Personal events impact music. Music reflects emotions. Emotions was then what hooked us, so our focus become how to analyze music based on emotions.

Since scholars suggest to cluster emotions into five primary categories (as with colors), namely joy, anger, sadness, fear and disgust. With this as our base, our product was born: LYMO (lyrics and emotions) – a database fueled by IBM Watson and Google Nngram with 100k+ songs from 1.8k artists, spanning 70 years and 11 genres.

Through big data analysis of text snippets, LYMO allows us to compile playlists to suit any emotion – the most joyful, saddest, or angriest playlists you like… Did you know that in rough times (e.g. Vietnam war) music tended to be more joyful? Think Barry White.

Did you know that over time, and especially in recent years, songs became more alike and emotionally less diverse? Think commercial music on the radio. Did you know that in the 80s when rap music rose it tended to be very angry? Think Gang Starr. Did you also know though that later in the 00s rappers showed some love and the genre became less angry? Think Wu-Tang Clan.

So, obviously, all kinds of questions can be answered by analyzing music through lyrics and their expressed emotions. Now with LYMO, take any sort of text snippet, analyze its emotions and match it with the emotional profile of a song: You have found the perfect match!

If you would like to learn more about LYMO and its potential, meet us at the MUSIC ANALYTICS station at our dataX event, on April 26, 2018.

There are no comments yet.

Leave a Comment

Required fields are marked with *. Your email address will never be published. Icons are imported via Gravatar. Please be polite and don’t spam. If it’s you first comment, it might have to go through moderation first.