From music audio to chord tablature: Teaching deep convolutional networks toplay guitar

Humphrey, Eric J.; Bello, Juan Pablo

doi:10.1109/icassp.2014.6854952

Cited by 20 publications

(9 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Boulanger-Lewandowski et al [11] train a recurrent neural network to produce chord classifications using input of PCA-whitened magnitude DFT. In a similar direction, Humphrey and Bello [32] build a DNN that maps input spectrogram features to guitar-specific fingerings of chords.…”

Section: A Overviewmentioning

confidence: 99%

Deep Learning and Music Adversaries

Kereliuk

Sturm

Larsen

2015

IEEE Trans. Multimedia

114

View full text Add to dashboard Cite

Abstract-An adversary is essentially an algorithm intent on making a classification system perform in some particular way given an input, e.g., increase the probability of a false negative. Recent work builds adversaries for deep learning systems applied to image object recognition, which exploits the parameters of the system to find the minimal perturbation of the input image such that the network misclassifies it with high confidence. We adapt this approach to construct and deploy an adversary of deep learning systems applied to music content analysis. In our case, however, the input to the systems is magnitude spectral frames, which requires special care in order to produce valid input audio signals from network-derived perturbations. For two different train-test partitionings of two benchmark datasets, and two different deep architectures, we find that this adversary is very effective in defeating the resulting systems. We find the convolutional networks are more robust, however, compared with systems based on a majority vote over individually classified audio frames. Furthermore, we integrate the adversary into the training of new deep systems, but do not find that this improves their resilience against the same adversary.

show abstract

Section: A Overviewmentioning

confidence: 99%

Deep Learning and Music Adversaries

Kereliuk

Sturm

Larsen

2015

IEEE Trans. Multimedia

114

View full text Add to dashboard Cite

show abstract

“…The data set included examples from the string (bowed and plucked), woodwind (single, double, and air reed), and brass families. Eric J. Henry et al [7] proposed a model that can yield representations for the chords that require minimal prior knowledge to interpret. The model has been developed to address both challenges by modeling the physical constraints of a guitar to produce human-readable representations of music audio, i.e.…”

Section: Literature Reviewmentioning

confidence: 99%

Impulse Response Modeling of the Box Shaped Acoustic Guitar

Atre¹,

Apte²

2022

Music in Health and Diseases

View full text Add to dashboard Cite

Music is the pulse of human lives and is an amazing tool to relieve and re-live. And when it comes to the signal processing, impulse is the pulse of the researchers. The work presented here is focused on impulse response modeling of noted produced by box shaped acoustic guitar. The impulse response is very fundamental behavior of any system. The music note is the convolution of the impulse response and the excitation signal of that guitar. The frequency of the generated music note follows the octave rule. The octave rule can be checked for impulse responses as well. If the excitation signal and impulse response are separated, then an impulse response of a single fret can be used to generate the impulse responses of other frets. Here the music notes are analyzed and synthesized on the basis of the plucking style and plucking expression of the guitar-player. If the impulse response of the musical instrument is known, the output music note can be synthesized in an unusual manner. Researchers have been able to estimate the impulse response by breaking the string of the guitar. Estimating the impulse response from the recorded music notes is possible using the methodology of cepstral domain window. By means of the Adaptive Cepstral Domain Window (ACDW) the author estimated the impulse response of guitar notes. The work has been further extended towards the classification of synthesized notes for plucking style and plucking expression using Neural Network and Machine Learning algorithms.

show abstract

“…Early works relied on MFCC input to reduce computation [62], [63] for genre classification. Many works have been then introduced based on time-frequency representations e.g., CQT for chord recognition [45], guitar chord recognition [46], genre classification [116], transcription [96], melspectrogram for boundary detection [89], onset detection [90], hit song prediction [118], similarity learning [68], instrument recognition [39], music tagging [26], [15], [17], [59], and STFT for boundary detection [36], vocal separation [100], and vocal detection [88]. One-dimensional CNN for raw audio input is used for music tagging [26], [60], synthesising singing voice [9], polyphonic music [112], and instruments [29].…”

Section: Convolutional Layers and Musicmentioning

confidence: 99%

“…For convnets, the depth is increasing in MIR as well as other domains. For example, networks for music tagging, boundary detection, and chord recognition in 2014 used 2layer convnet [26], [90], [46], but recent research often uses 5 or more convolutional layers [15,68,52].…”

Section: Depth Of Networkmentioning

confidence: 99%

A Tutorial on Deep Learning for Music Information Retrieval

Choi,

Fazekas,

Cho

et al. 2017

Preprint

View full text Add to dashboard Cite

Following their success in Computer Vision and other areas, deep learning techniques have recently become widely adopted in Music Information Retrieval (MIR) research. However, the majority of works aim to adopt and assess methods that have been shown to be effective in other domains, while there is still a great need for more original research focusing on music primarily and utilising musical knowledge and insight. The goal of this paper is to boost the interest of beginners by providing a comprehensive tutorial and reducing the barriers to entry into deep learning for MIR. We lay out the basic principles and review prominent works in this hard to navigate field. We then outline the network structures that have been successful in MIR problems and facilitate the selection of building blocks for the problems at hand. Finally, guidelines for new tasks and some advanced topics in deep learning are discussed to stimulate new research in this fascinating field.

show abstract

From music audio to chord tablature: Teaching deep convolutional networks toplay guitar

Cited by 20 publications

References 7 publications

Deep Learning and Music Adversaries

Deep Learning and Music Adversaries

Impulse Response Modeling of the Box Shaped Acoustic Guitar

A Tutorial on Deep Learning for Music Information Retrieval

Contact Info

Product

Resources

About