An Enhanced Musical Instrument Classification using Deep Convolutional Neural Network

Retrieval of musical information from musical databases is a major challenging issue in a digital world. Therefore, it is necessary to develop an efficient tool for retrieving the musical information. Musical instrument classification plays a major role for retrieving the information from musical database. In order to retrieve the musical instrument efficiently, an enhanced musical instrument classification algorithm using deep Convolutional Neural Network is proposed in this paper. The proposed algorithm consists of convolutional layers interleaved with two pooling functions followed by two fully interconnected layers. There are sixteen instruments from different instrument families are taken for evaluating the performance of proposed algorithm. The experimental result shows that the proposed algorithm recognizes the instruments significantly and achieves the greater accuracy than existing algorithm

Various Approaches in Musical Instrument Identification

In the music information retrieval (MIR) field, it is highly desirable to know what instruments are used in an audio sample. Musical instrument classification is one of the sub domains of music information retrieval. Many researchers have presented different approaches for identifying western instruments and those approaches proved to be good for instrument identification. In this article, we have presented work done by the various authors to identify musical instrument using various approaches such sparse based representation, bio-inspired hierarchical model, joint modelling, Bayesian networks, neural networks, convolution neural networks, individual partials, clustering, and segmentation.

Hybridisation of Mel Frequency Cepstral Coefficient and Higher Order Spectral Features for Musical Instruments Classification

This paper presents the classification of musical instruments using Mel Frequency Cepstral Coefficients (MFCC) and Higher Order Spectral features. MFCC, cepstral, temporal, spectral, and timbral features have been widely used in the task of musical instrument classification. As music sound signal is generated using non-linear dynamics, non-linearity and non-Gaussianity of the musical instruments are important features which have not been considered in the past. In this paper, hybridisation of MFCC and Higher Order Spectral (HOS) based features have been used in the task of musical instrument classification. HOS-based features have been used to provide instrument specific information such as non-Gaussianity and non-linearity of the musical instruments. The extracted features have been presented to Counter Propagation Neural Network (CPNN) to identify the instruments and their family. For experimentation, isolated sounds of 19 musical instruments have been used from McGill University Master Sample (MUMS) sound database. The proposed features show the significant improvement in the classification accuracy of the system.