Listening Machine Turns Tweets To Music

A website known as the Listening Machine has pledged to turn the next six months of tweets by 500 unnamed users into a "live soundtrack to the thoughts, opinions, feelings and conversations of the UK's population, as played out on Twitter."

Created by programmer Daniel Jones and composer Peter Gregson, the Listening Machine has already begun its translation, creating a strange mix of plinky plonky sounds with the background sound of conversation and other real world noises. It works by analysing tweets for their sentiment, rate and topic, using the categories they fit into to generate music with varied instruments.

Sentiment swings between positive, negative and neutral, while the rate is calculated at tweets per minute - presumably working out somehow into beats per minute for the musical track. Some of the topics that a tweet can relate to, thereby selecting certain instruments and mood, include: Arts, Business, Education, Health, Politics, Science, Sport and Technology. The most popular at the time of writing is Science, with Health and Sport coming in second and third respectively.

While some segments are pre-recorded and designed to show up with certain phrases or sentiments, the words themselves in tweets are often translated into specific notes and melodies.

"Honestly, producing six months worth of potential music was a nightmare," said Gregson. "Clearly we didn't have the time (nor the inclination) to write, record, edit and mix six months worth of music, so we had to be creative and think hard about how to map music cells to words, letters, consonants, vowels, sentiment and so on. The hardest part was itemising all the bits we would need and develop ways that, even if every single sound played all at once, it wouldn't sound crap! It's like painting -- if you use every colour you've got, everything looks brown."

The project is set to run until October 2012 and has been funded by the BBC and the Arts Council.

Source: Wired