Million song search: web interface for semantic music search using musical word embedding

We present a web interface for large-scale semantic search using a musically customized word embedding in the backend. The musical word embedding represents artist entities, track entities, tags, and ordinary words in a single vector space. It is learned based on the affinity between the words and the entities using a wide spectrum of text data including Wikipedia, music review, and music tags. The system can allows users to type a query within 9.8M vocabulary words in the musical word embedding. It also supports a multi-query blending function using a semantic averaging of the queries to provide more refined search.