Musical instrument sound classification with deep convolutional neural network using feature fusion approach
This work addresses the problem of accurately classifying musical instruments for audio processing applications, representing an incremental improvement by enhancing feature extraction with phase information.
The paper tackles musical instrument sound classification by proposing a feature fusion approach that combines spectrogram images with novel multiresolution recurrence plots (MRPs) to capture phase information, using a deep convolutional neural network. The result shows improved performance over methods using only spectrograms or traditional handcrafted features, though specific numbers are not provided.
A new musical instrument classification method using convolutional neural networks (CNNs) is presented in this paper. Unlike the traditional methods, we investigated a scheme for classifying musical instruments using the learned features from CNNs. To create the learned features from CNNs, we not only used a conventional spectrogram image, but also proposed multiresolution recurrence plots (MRPs) that contain the phase information of a raw input signal. Consequently, we fed the characteristic timbre of the particular instrument into a neural network, which cannot be extracted using a phase-blinded representations such as a spectrogram. By combining our proposed MRPs and spectrogram images with a multi-column network, the performance of our proposed classifier system improves over a system that uses only a spectrogram. Furthermore, the proposed classifier also outperforms the baseline result from traditional handcrafted features and classifiers.