Melody Extraction from Polyphonic Music Using Deep Neural Network: A Literature Survey

Abstract

Abstract: Melody extraction plays an important role in the field of Music Information Retrieval (MIR). It has emerged as one of the active research problems in the MIR applications. Nowadays, the music providers have to facilitate searching of music based on their contents or recommend music based on user’s interest having similar contents. Melody extraction is necessary to fulfil these user-interest driven searching and recommendation. The main objective of melody extraction is to generate a sequence of frequency parallel to the pitch of the dominant melody from the music. Numerous approaches have been developed for melody extraction. This survey gives an overview of Deep Neural Network (DNN) based melody extraction approaches. These approaches primarily focus on melody extraction from polyphonic music. In addition, this survey also discusses the commonly used datasets and the performance measures that are used to evaluate the performance of melody extraction approaches. The pros and cons of algorithms are highlighted based on the type of data used for training and testing. Future perspective of DNN based approaches is also discussed.

Keywords: Music Information Retrieval (MIR), Deep Neural Network (DNN), Long Short Term Memory (LSTM), Deep Convolutional Neural Network (DCNN), Deep Harmonic Neural Networks (DHNN)

Cite this Article: Ranjeet Kumar, Anupam Biswas, Pinki Roy. Melody Extraction from Polyphonic Music Using Deep Neural Network: A Literature Survey. Journal of Software Engineering Tools & Technology Trends. 2019; 6(3): 16–21p.