ISP - Intelligent song picker
Challenge
Have you ever felt overwhelmed when choosing a song from Spotify?
I tried to create a prototype that relieves users of the everyday decision of what music they want to listen to.
Concept, prototyping and development of a web-application using image analysis with machine learning and an intelligent song selection process to automate that mundane decision.
Link
Insights
ISP leverages modern machine learning algorithms to counteract decision fatigue. It provides users with a mood and environment complimenting song based on a picture of themselves.
In addition to parameters such as color, brightness and contrast, facial and emotion recognition are used to evaluate the user's setting.
Based on the image processing, an intelligent algorithm then uses the Spotify API to select a personalized song that matches the mood and environment.
More detailed insights into the code, the use of web workers, Spotify API and Tensorflow.js models are available in the public repository.
I even wrote a whole paper about this project.