Artify

For a given song, we fetched similar songs and performed topic modeling using LDA to extract common themes from the lyrics. These themes are representative of the emotion and mood of the song. We used this information to find art that match the extracted themes. We then used a machine learning technique called artistic neural transfer to apply the extracted style on a user’s profile picture to reflect the user’s current state of mind.
Artify is a social sharing feature that applies artistic filters to images based on emotions in music. We use the Spotify API to get the song that a user is listening to and apply a natural language processing based pipeline to retrieve emotions from the lyrics. The emotions are compared to tags associated with a stylistic template, and a template is picked for style transfer to the user’s website.