JamSearch

• 

  Find your jam in our kitchen ! Isn't that aMAISing ?

Inspiration

We all like discovering new music right? So why not combine it with ML and create a song recommendation system?

What it does

You can search for songs you like; save them and predict ! You will then have a selection of songs just for you based on the songs you saved

How we built it

We used a Kaggle dataset that had around 200k songs; and then used cosine similarity to predict songs close to a mean one (=the liked songs).

Challenges we ran into

We mostly had trouble coming up with the correct model for this project.

Accomplishments that we're proud of

To be honest, we are proud of anything that works and doesn't throw errors at us :) We also like the fact that everyone gets to discover new music from something we implemented.

What we learned

We learn how to implement unsupervised learning and what it entails.

What's next for JamSearch

We would like to take things to the next level by tweaking the web-app as well as directly accessing a user's spotify library to get their liked songs to get even more accurate results.

Built With

• flask
• javascript
• kaggle
• python
• react
• spotify