What if you could take a tweet and turn it into a playlist?
What it does
We made a chrome extension that takes textual input from the browser (tweets, facebook posts, paragraphs from news articles, etc - basically any text on any website) and use this as a seed to generate playlist.
To do so we are considering several factors - the most important of which is the emotional analysis and valence analysis of the text. We are using this to identify songs that best match the 'mood' of the text. As an easter egg, if you look at the first letter of all the songs in the playlist, it spells out the seed text.
How we built it
We used Bottle with Python in the backed, Spotify's API to get the songs and to create the playlist, IBM Watson to analyze on the text, and jQuery and Matterialize for the front end. Further we created a Chrome extension to provide easy access to our project.
Challenges we ran into
- Learning how to make a chrome extension. It was quite difficult to figure out how to manage the scope of our code, as it was not what I was used to.
- Adding a custom filter to a search API that doesn't support filters (only keywords) and still being efficient.
- Tried to implement sentiment analysis on my own, but then reverted to IBM when our results weren't that good.
Accomplishments that we're proud of
We made something that we could actually use IRL. Usually we just do silly projects in which we try out new tech. This time we did something new, and also useable.
What's next for getREC'd
Hopefully we can improve the genre selection.