Real-time black-box modelling with recurrent neural networks

Research output: Chapter in Book/Report/Conference proceedingConference contributionScientificpeer-review

Standard

Real-time black-box modelling with recurrent neural networks. / Wright, Alec; Damskägg, Eero-Pekka; Välimäki, Vesa.

Proceedings of the International Conference on Digital Audio Effects. University of Birmingham, 2019. (Proceedings of the International Conference on Digital Audio Effects).

Research output: Chapter in Book/Report/Conference proceedingConference contributionScientificpeer-review

Harvard

Wright, A, Damskägg, E-P & Välimäki, V 2019, Real-time black-box modelling with recurrent neural networks. in Proceedings of the International Conference on Digital Audio Effects. Proceedings of the International Conference on Digital Audio Effects, University of Birmingham, International Conference on Digital Audio Effects, Birmingham, United Kingdom, 02/09/2019.

APA

Wright, A., Damskägg, E-P., & Välimäki, V. (2019). Real-time black-box modelling with recurrent neural networks. In Proceedings of the International Conference on Digital Audio Effects (Proceedings of the International Conference on Digital Audio Effects). University of Birmingham.

Vancouver

Wright A, Damskägg E-P, Välimäki V. Real-time black-box modelling with recurrent neural networks. In Proceedings of the International Conference on Digital Audio Effects. University of Birmingham. 2019. (Proceedings of the International Conference on Digital Audio Effects).

Author

Wright, Alec ; Damskägg, Eero-Pekka ; Välimäki, Vesa. / Real-time black-box modelling with recurrent neural networks. Proceedings of the International Conference on Digital Audio Effects. University of Birmingham, 2019. (Proceedings of the International Conference on Digital Audio Effects).

Bibtex - Download

@inproceedings{769f627fa4fe49569bd207f6b1d32dc3,
title = "Real-time black-box modelling with recurrent neural networks",
abstract = "This paper proposes to use a recurrent neural network for black-box modelling of nonlinear audio systems, such as tube amplifiers and distortion pedals. As a recurrent unit structure, we test both Long Short-Term Memory and a Gated Recurrent Unit. We compare the proposed neural network with a WaveNet-style deep neural network, which has been suggested previously for tube amplifier modelling. The neural networks are trained with several minutes of guitar and bass recordings, which have been passed through the devices to be modelled. A real-time audio plugin implementing the proposed networks has been developed in the JUCE framework. It is shown that the recurrent neural networks achieve similar accuracy to the WaveNet model, while requiring significantly less processing power to run. The Long Short-Term Memory recurrent unit is also found to outperform the Gated Recurrent Unit overall. The proposed neural network is an important step forward in computationally efficient yet accurate emulation of tube amplifiers and distortion pedals.",
author = "Alec Wright and Eero-Pekka Damsk{\"a}gg and Vesa V{\"a}lim{\"a}ki",
year = "2019",
month = "9",
day = "2",
language = "English",
series = "Proceedings of the International Conference on Digital Audio Effects",
publisher = "University of Birmingham",
booktitle = "Proceedings of the International Conference on Digital Audio Effects",

}

RIS - Download

TY - GEN

T1 - Real-time black-box modelling with recurrent neural networks

AU - Wright, Alec

AU - Damskägg, Eero-Pekka

AU - Välimäki, Vesa

PY - 2019/9/2

Y1 - 2019/9/2

N2 - This paper proposes to use a recurrent neural network for black-box modelling of nonlinear audio systems, such as tube amplifiers and distortion pedals. As a recurrent unit structure, we test both Long Short-Term Memory and a Gated Recurrent Unit. We compare the proposed neural network with a WaveNet-style deep neural network, which has been suggested previously for tube amplifier modelling. The neural networks are trained with several minutes of guitar and bass recordings, which have been passed through the devices to be modelled. A real-time audio plugin implementing the proposed networks has been developed in the JUCE framework. It is shown that the recurrent neural networks achieve similar accuracy to the WaveNet model, while requiring significantly less processing power to run. The Long Short-Term Memory recurrent unit is also found to outperform the Gated Recurrent Unit overall. The proposed neural network is an important step forward in computationally efficient yet accurate emulation of tube amplifiers and distortion pedals.

AB - This paper proposes to use a recurrent neural network for black-box modelling of nonlinear audio systems, such as tube amplifiers and distortion pedals. As a recurrent unit structure, we test both Long Short-Term Memory and a Gated Recurrent Unit. We compare the proposed neural network with a WaveNet-style deep neural network, which has been suggested previously for tube amplifier modelling. The neural networks are trained with several minutes of guitar and bass recordings, which have been passed through the devices to be modelled. A real-time audio plugin implementing the proposed networks has been developed in the JUCE framework. It is shown that the recurrent neural networks achieve similar accuracy to the WaveNet model, while requiring significantly less processing power to run. The Long Short-Term Memory recurrent unit is also found to outperform the Gated Recurrent Unit overall. The proposed neural network is an important step forward in computationally efficient yet accurate emulation of tube amplifiers and distortion pedals.

M3 - Conference contribution

T3 - Proceedings of the International Conference on Digital Audio Effects

BT - Proceedings of the International Conference on Digital Audio Effects

PB - University of Birmingham

ER -

ID: 36768416