The majority of musicians and composers in United States are independent and do not have any access to the major distributional channels in music industry — they don't reach their potential fans! When they can use a PR firm it is extremely costly (5000 dollars an album). Further each music reviewer is spammed mails containing music links and reviewers spend several hours finding the right song.
What it does
Our app solves this problem-it matches musicians to reviewers which are most likely to publish there song. It asks users to submit the SoundCloud link to their song, then it runs this song through Echonest API, which makes it comparable to other songs already published by different bloggers, and finds the most the perfect match with the reviewer! TrackRipple is time efficient and cheap alternative to expensive and dominant producing agencies and PR firms!
How we built it
Our first step was collecting the database of blogs and songs they were reviewing by scraping the music blog directory available from link. Then we built a backend that populated our productional database and ran the matching algorithm to find blogs which were most likely to like the new song submitted to our service. Finally, we deployed Flask-driven fronted that asked users to submit their SoundCloud links for review with their contact information.
Challenges we ran into
Matching algorithm was based on testing a statistical hypothesis on whether the song is a good match for a certain blog. Each song was represented as a 8-dimensional vector of numeric parameters, however as it was revealed during the first tests, each parameter does not contribute equally to defining the difference between songs. Thus, proper weight assignment is our next challenge.
Accomplishments that we're proud of
We can break down songs into parameters and can clearly see that one blog publishes songs which is parametrically clustered together. Now its just a matter of weight assignment in establishing a clear match for every new song.
What's next for TrackRipple
Several possibilities like using this platform as a marketplace for musicians, collaborating with sound markets (talks on the way with scoreAscore), or simply using the algorithm to similar applications in music industry.
Check out our video demo: link