News

Deep learning can fool listeners by imitating any guitar amplifier

A study from the Aalto Acoustics Lab demonstrates that digital simulations of guitar amplifiers can sound just like the real thing. The implications are that as the software models continue to improve, they can replace traditional analogue guitar amplifiers, which are bulky, fragile and expensive.
Testing guitar signals at the Aalto Acoustics Lab
Testing guitar signals at the Acoustics Lab

Many popular guitar amplifiers and distortion effects are based on analogue circuitry. To achieve the desired distortion of the guitar signal, these circuits use nonlinear components, such as vacuum tubes, diodes, or transistors. As music production becomes increasingly digitised, the demand for faithful digital emulations of analogue audio effects is increasing.  

Professor Vesa Välimäki explains that this is an exciting development in deep learning, ‘Deep neural networks for guitar distortion modelling has been tested before, but this is the first time, where blind-test listeners couldn’t tell the difference between a recording and a fake distorted guitar sound! This is akin to when the computer first learned to play chess’. 

The main objective of the field of Virtual Analog (VA) modelling is to create digital emulations of these analogue systems which will allow bulky, expensive and fragile analogue equipment to be replaced by software plugins that can be used on a modern desktop or laptop computer.  

A specific amplifier’s circuitry can be accurately simulated using circuit modelling techniques, but the result is often a model that is too computationally demanding for real-time processing. Additionally, a new model has to be created for each amplifier being modelled, and the process is labour intensive.

A basic schematic of the guitar signal between the neural net and amplifier

An alternative approach for VA modelling is “black-box” modelling. Black-box modelling is based on measuring the circuit’s response to some input signals and creating a model which replicates the observed input-output mapping. The study from which these results came, was based on the WaveNet convolutional neural network. 

The digital amplifier model is created using a deep neural network. Audio is recorded from a ’target ’guitar amplifier, and this audio is used to train the deep neural network to simulate that guitar amplifier. 

Alec Wright, a doctoral student, focusing on audio processing using deep learning says,’ The tests were conducted to validate the performance of models emulating either the Blackstar HT5 Metal or Mesa Boogie Express 5:50+ tube amplifiers. The models were created with a focus on real-time performance, and all of them can be run in real-time on a desktop computer’. 

All of this means that in the near future, all a guitarist will need to do is plug into their laptop that is running the deep neural plugin, and a thoroughly convincing vintage guitar amp sound will come from the speakers.  

It remains to be seen if guitar amplifier purists will be willing to part with their beloved rigs, but this innovation paves the way for any audio enthusiast to digitally get the desired guitar sound, whether it be a Marshall, Orange, Fender, or anything else, on the road or in the studio.  

Read the paper here: https://www.mdpi.com/2076-3417/10/3/766

More information

Alec Wright

Alec Wright

Doctoral Researcher
T405 Dept. Signal Processing and Acoustics

Study Acoustics and Audio Technology at Aalto University

Three Aalto University students working in a acoustics laboratory

Acoustics and Audio Technology - Computer, Communication and Information Sciences, Master of Science (Technology)

The major in Acoustics and Audio Technology equips students with a fundamental understanding of human hearing, audio perception, and physics of sound. The skills they acquire enable, for example, reducing noise pollution, planning harmonic environments and designing coherent sound experiences.

Study options
  • Published:
  • Updated:
Share
URL copied!

Read more news

Professor Jens Schmidt opening the first Aalto Alumni Strategy Circle. Audience in the front.
Research & Art Published:

Aalto Alumni Strategy Circle shared best practices in mergers and acquisitions

The Aalto Alumni Strategy Circle brings together strategy professionals and business leaders to learn and exchange ideas about the most relevant current topics in strategy.
Food Futures, ATARCA
Cooperation, Press releases, Research & Art Published:

Learning by doing: Solving the Tragedy of the Commons one Meal at a time

Food Futures, part of ATARCA, Aalto University, in collaboration with Global Politics and Communication, University of Helsinki, present the video Learning by Doing: Solving the Tragedy of the Commons one Meal at a time.
Anne Sunikka Dipolissa tietokoneen ääressä.
Research & Art Published:

Quality together – University supports open access in science and research

Support, training and information available for researchers on open science practices
A solar panel with blue sky and a graphic showing perovskite material
Research & Art Published:

Machine learning boosts the discovery of new perovskite solar cell materials

Researchers from CEST group apply machine learning to help identifying suitable solar cell materials