MIC_FuzzyNET: Fuzzy Integral Based Ensemble for Automatic Classification of Musical Instruments From Audio Signals

Music has been an integral part of the history humankind with theories suggesting it is more antediluvian than speech itself. ordered succession tones and harmonies that produce sounds characterized by melody rhythm. Our paper proposes ensemble deep learning musical instrument classification (MIC) framework, named as MIC_FuzzyNET model which aims to classify dominant instruments present in clips. Firstly, data converted three different spectrograms: Constant Q-Transform, Semitone Spectrogram, Mel are then stacked form 3 channel 2D data. This spectrogram fed transfer models namely, EfficientNetV2 ResNet18 output preliminary scores. A fuzzy rank finally employed assigns classifier ranks, on testing achieve final enhanced scores reduces error biases for constituent CNN architectures. proposed framework evaluated Persian Classical Instrument Recognition (PCMIR) dataset Musical Audio Signals (IRMAS) dataset. It achieved considerably high accuracy, making our a robust MIC model.