Bored by the redundant recommendations by algorithms
on music platforms I tried to find a new way of
discovering new artists and tracks: Instead of
having an algorithm analyse the overall plays and
try to predict the next video a person might play, I
wanted to use an algorithm sorely to connect two
persons with a similar music taste.
In my perception, human based selection had some
advantages when it comes to finding new music: I
noticed, that algorithms tend to run in circles,
since they didn't take into account already known
tracks. Further, algorithms are not able to
understand cultural context, so their
recommendations are quite narrow and rarely result
in unrelated, yet interesting results.
The aspect that made pursuit this project was the
idea that the user could follow his own interest by
saving tracks for himself, yet creating a value for
other users.
To test this, I built a fully functional prototype
with a working database for user management and
comparing different playlists,
Soundcloud
and
Youtube API
requests to gather the content, and a working user
interface based on the
Meteor framework.
Since I did this project in a course focused on
business, I also analysed the potential and
weaknesses of marketing the platform. You can find
the results in the attachments.
Screenshots taken from functional application. Data provided by Soundclound and Youtube API.