Musical Instrument Identification using Supervised Learning

This research article has used supervised machine learning algorithms for identifying musical instruments. For every sound sample, there will be feature vectors. This machine learning model utilize the feature vectors for representing the numbers or symbols which are also known as features, so that they can be analyzed easily. Generally, Machine learning algorithms require a numerical representation of objects so that processing of data in numerical fashion will be more accurate and easier and also feature vector helps in statistical analysis in a better way. Every problem represented in a feature makes it understand deeper. For example, it is known that the colours are a combination of red, green and blue (RGB), every colour has a specific RGB value here values are also known as features. Vector defines it as a collection or set of numbers like a matrix whereas the feature is something which defines the property of an object representing numerically, putting together both features and vectors gives feature vector, a feature vector is something which defines the property of objects represented in numerals. Feature vectors are used in classification problems. This research work has used a very small data set containing a total of 600 audio files from six musical instruments such as violin, guitar, flute, saxophone, oboe, trumpet, cello [10]. Further, this research work has achieved a high classification accuracy on the testing data with help of a support vector machine using RBF kernel and k nearest neighbour.