Deep Neural Networks: A Case Study for Music Genre Classification

Rajanna, Arjun Raj; Aryafar, Kamelia; Shokoufandeh, Ali; Ptucha, Raymond

doi:10.1109/icmla.2015.160

Cited by 33 publications

(6 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…From its inception at Bell labs in the 1950s with the 'Audrey' system, capable of recognising spoken digits (Davis et al, 1952), through the considerable advancements during the 1980s associated with the use of hidden Markov models (Hansen & Hasan, 2015), and to the recent deep learning revolution (Hinton et al, 2012), ASR technologies have now matured to the point where they are embedded in everyday technologies, for example, SIRI ™ , CORTANA ™ , and ALEXA ™ . A similar transforming effect has recently occurred through deep learning, in terms of the immense increase in recognition accuracy and robustness in music analysis (e. g., Coutinho et al, 2014;Rajanna et al, 2015;Sigtia et al, 2016), and for the recognition of acoustic scenes and the detection of specific audio events (Mesaros et al, 2018).…”

Section: Introductionmentioning

confidence: 69%

New Avenues in Audio Intelligence: Towards Holistic Real-life Audio Understanding

et al. 2021

View full text Add to dashboard Cite

Computer audition (i.e., intelligent audio) has made great strides in recent years; however, it is still far from achieving holistic hearing abilities, which more appropriately mimic human-like understanding. Within an audio scene, a human listener is quickly able to interpret layers of sound at a single time-point, with each layer varying in characteristics such as location, state, and trait. Currently, integrated machine listening approaches, on the other hand, will mainly recognise only single events. In this context, this contribution aims to provide key insights and approaches, which can be applied in computer audition to achieve the goal of a more holistic intelligent understanding system, as well as identifying challenges in reaching this goal. We firstly summarise the state-of-the-art in traditional signal-processing-based audio pre-processing and feature representation, as well as automated learning such as by deep neural networks. This concerns, in particular, audio interpretation, decomposition, understanding, as well as ontologisation. We then present an agent-based approach for integrating these concepts as a holistic audio understanding system. Based on this, concluding, avenues are given towards reaching the ambitious goal of ‘holistic human-parity’ machine listening abilities.

show abstract

Section: Introductionmentioning

confidence: 69%

New Avenues in Audio Intelligence: Towards Holistic Real-life Audio Understanding

et al. 2021

View full text Add to dashboard Cite

show abstract

“…Literature shows that providing complex features, e.g., generalized cross-correlation, spectral power, and phase spectrum, as the input significantly improves the classification or regression performance of a DNN compared to raw timedomain input (Rajanna et al, 2015). Any input stream corruption pollutes the generated features and causes poor DNN inference performance.…”

Section: Introductionmentioning

confidence: 99%

RecNet: Early Attention Guided Feature Recovery

Biswas¹,

Islam²

2023

Preprint

View full text Add to dashboard Cite

Uncertainty in sensors results in corrupted input streams and hinders the performance of Deep Neural Networks (DNN), which focus on deducing information from data. However, for sensors with multiple input streams, the relevant information among the streams correlates and hence contains mutual information. This paper utilizes this opportunity to recover the perturbed information due to corrupted input streams. We propose RecNet, which estimates the information entropy at every element of the input feature to the network and interpolates the missing information in the input feature matrix. Finally, using the estimated information entropy and interpolated data, we introduce a novel guided replacement procedure to recover the complete information that is the input to the downstream DNN task. We evaluate the proposed algorithm on a sound event detection and localization application where audio streams from the microphone array are corrupted. We have recovered the performance drop due to the corrupted input stream and reduced the localization error with non-corrupted input streams.

show abstract

“…In recent years, the new classifiers have also been used for musical genre classification. Deep convolutional neural networks (CNN) have been used for the classification of musical genres [20,21].…”

Section: Introductionmentioning

confidence: 99%

A New Genre Classification with the Colors of Music

Durdağ

Erdoğmuş²

2019

Sakarya University Journal of Computer and Information Sciences

View full text Add to dashboard Cite

The aim of this study is to bring a new perspective for the classification of the songs, revealing of the colors of music. The first effort is to transform songs into images. The colorful images have been attained with Short time Fourier transform, discrete cosine transform and time to spatial transformation and some extra processing. It has been observed that the images of different music genres obtained with the same method have different colors. But some of them have similar colors and patterns, which making difficult to classify. Pre-trained deep convolutional network have been trained with these images. For five Turkish musical genres, nearly up to 60% classification accuracy has been achieved and for ten musical genre of a benchmark musical dataset nearly up to 54% classification accuracy has been achieved. In future studies, it has been planned to create the images using timbral texture and rhythmic contents, for increasing the accuracy.

show abstract

Deep Neural Networks: A Case Study for Music Genre Classification

Cited by 33 publications

References 28 publications

New Avenues in Audio Intelligence: Towards Holistic Real-life Audio Understanding

New Avenues in Audio Intelligence: Towards Holistic Real-life Audio Understanding

RecNet: Early Attention Guided Feature Recovery

A New Genre Classification with the Colors of Music

Contact Info

Product

Resources

About