Musical Word Embedding: Bridging the Gap between Listening Contexts and Music