Moving Beyond Feature Design: Deep Architectures and Automatic Feature Learning in Music Informatics

By Eric J. Humphrey, Juan Pablo Bello, Yann LeCun. 2012

154 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

1. Li Deng
2. Yoshua Bengio
3. Dong Yu
4. Yann LeCun
5. Colin Raffel
6. Jianfeng Gao
7. Xavier Bresson
8. Kyunghyun Cho

Citations in the corpus

• Deep Learning: Methods and Applications (2014, background): “In the general field of audio and music processing, the impacted areas by deep learning include mainly music signal processing and music information retrieval [15, 22, 141, 177, 178, 179, 319].”
• A tutorial survey of architectures, algorithms, and applications for deep learning (2014, background): “The impacted areas include mainly music signal processing and music information retrieval [49, 185–187].”
• Feature learning and deep architectures: new directions for music informatics (2013, background): “2012)? Are we considering all possible solutions (Humphrey et al. 2012)?”
• Precision Learning: Towards Use of Known Operators in Neural Networks (2017, background): “Interestingly, emerging networks share similarities with known operators [7], [8].”
• Improving Content-based and Hybrid Music Recommendation using Deep Learning (2014, background): “In [36, 37], Humphrey et al. proposed that the traditional two-stage machine learning process — feature extraction and classi-fication/regression — should be conducted simultaneously.”
• End-to-end learning for music audio (2014, background): “Lately, feature learning has also been receiving more attention from the MIR community [3].”
• FMA: A Dataset for Music Analysis (2016, background): “While those approaches have fallen behind learning from higher-level representations such as spectrograms [5], a greater exploration of the design space will hopefully provide alternatives to solving MIR challenges [16].”
• A Software Framework for Musical Data Augmentation (2015, methodology): “As an alternative to custom feature design, some authors advocate learning or optimizing features directly from the data [11].”
• Speech emotion recognition using convolutional and Recurrent Neural Networks (2016, unknown): “Because of this, the traditional trend in speech/audio information retrieval is to focus on the use of powerful strategies for semantic analysis, often relying on model selection to optimize the results [2].”