• Create an user attributed playlist

• Use music databases to find suggestions for similar songs

​

After completing our main goals, our ultimate ambition was to make an user attributed playlist. After we organize songs according to their key and tempo similarities, our next step was to add user preferences. In the user data, the user would go through a playlist, and rate from a scale of 1 to 5, lowest to highest respectively. Then, the user would also have the option to control the variance of the songs by choosing rarely, sometimes, or frequently.

 

From the data collected from the user and along with our playlist analysis, we were going to make an algorithm that suggested songs of a specific artist the user liked the most, or possibly a genre the user seemed to highly rate. To do this, we were going to use an existing database, “Last.fm” by Million Song Dataset [7]. From this existing database which has 584,897 songs with at least one similar song, 522,366 songs with unique tags, and data linking artist ID to the tags, we were going to create two smaller versions of this data, with perhaps 30 to 100 songs, for the user to go through for the user preferences. From this large set of data, we hoped to find similar songs that the user might like for a new favorite playlist.