Herd It’s blog begins
At UC San Diego’s Computer Audition Laboratory, we work on using machines to automatically analyze and understand music. Ph.D students, undergrads, professors and musicians all collaborate and use signal processing, machine learning, data mining and, now, human computation to discover everything we can about music.
In order for a computer to understand the difference between punk and metal or happy and sad music, the machine needs to learn about how each of these types of music sound. That means that we – human experts and music fans – need to teach the computers to understand these differences. Just like a child learns to tell opera from Obie Trice, we teach our computer audition algorithms to learn to recognize all types of music.
Herd It is the way that we collect data to train the machines. When you play Herd It, apart from listening to great music, playing fun games and connecting with other music fans, you are contributing examples of words that describe music. We use these tags as examples to train computer audition system so that it can listen to, understand and, ultimately, help us discover and enjoy every song ever recorded.
So, now that you know all about Herd It, go play!
www.herdit.org

Are you releasing the data to other researchers?
Joseph Turian
28 Sep 09 at 2:00 pm
Hi Joseph,
So far, Herd It’s data collection is in an early stage. Once we have collected a reasonable amount of data, we plan to validate and release it to the research community.
In the meantime, you can check out the CAL500 dataset that we have previously released as well as Edith Law’s MagnaTagATune dataset.
herdit
28 Sep 09 at 2:14 pm
Good luck on the HerdIt blog.
Daniel Kane
29 Sep 09 at 9:39 am