Self-learning music recommendation software.
Has this ever happened to you? You are relaxing and listening to a good song and all of a sudden your player selects a song that doesn’t fit into your current mood because you activated the shuffle mode. To solve this problem – similar to nowadays available commercial solutions like iTunes’ Genius or Spotify – we developed a piece of software that automatically analyses songs as you import them into your system (beats per minute, dynamics, meta-data) and fuel these information into a self-learning, trained neural network. Once the network contains the information it can recommend matching songs and also be trained to improve the quality of recommendations. This project was the final work during my time at the HTL-Ottakring.