Psychoacoustics of Machine Learning Amp Emulation Plugins

I recently presented my Master's thesis research on "Psychoacoustics of Machine Learning Amp Emulation Plugins" at the Harmonising Progress - Research in Music Production & Technology (HRMPT) symposium. This study examines how listeners perceive the similarity between machine learning and non-machine learning emulation amplifiers compared to the genuine Vox AC30 amplifier.

Key Findings:

• Machine learning emulation was perceived as most similar to the original amplifier in two out of three tested samples.

• In one sample, the machine learning emulation received a perfect median similarity rating of 100.00, significantly outperforming non-machine learning alternatives.

• Results showed strong statistical significance across all samples (p < 0.001).

• The study suggests that machine learning technologies can closely replicate the sound of traditional amplifiers, often surpassing non-machine learning emulations in perceived similarity.

Presenting research on machine learning amp emulation at HRMPT 2024

Methodology:

The study utilized a MUSHRA listening test, gathering data from a diverse group of participants including audio experts, trained musicians, and music technology students. It compared the original Vox AC30, its machine learning emulation using Neural Amp Modeler (NAM), and non-machine learning emulations using Waves GTR and IK Multimedia AmpliTube.

Implications: This research contributes to the ongoing evolution of digital audio tools and may significantly influence future developments in music production and sound engineering. It highlights the potential of machine learning in creating highly accurate digital emulations of classic audio equipment.

Engaged audience at the Arthur Sykes Rymer Auditorium during the presentation

Presentation Details:

The presentation took place on May 20, 2024, at the University of York's Arthur Sykes Rymer Auditorium as part of the HRMPT symposium. This event, organized by MA Music Production and Audio Cultures and MSc Audio and Music Technology students, showcases innovative projects spanning technology and creative arts.

For those interested in learning more about this study or discussing potential collaborations, please feel free to reach out.