A Novel Method to Detect Instrumental Music in a Large Scale Music Catalog

A large scale music catalog contains diverse types of sound recordings. In this paper, we present a methodology to identify instrumental music with high precision and high recall. Our method starts by separating a recording into a vocals track and a background track. Then, we process the vocals track with a singing voice detection model, to estimate the amount of singing voice in a song. Finally, we analyze the background track using a neural network trained to classify be-tween instrumental and non-instrumental sounds. We demonstrate the effectiveness of our approach on instrumental music detection through a comparative evaluation against several state-of-the-art models.