Polyphonic pitch recognition for guitar with neural networks

Thumbnail Image

Files

master_Jankowski_Antoni_2025.pdf (4.51 MB)

URL

Journal Title

Journal ISSN

Volume Title

School of Electrical Engineering | Master's thesis
Unless otherwise stated, all rights belong to the author. You may download, display and print this publication for Your own personal use. Commercial use is prohibited.

Authors

Date

2025-06-27

Department

Major/Subject

Acoustics and Audio Technology

Mcode

Degree programme

Master's Programme in Computer, Communication and Information Sciences

Language

en

Pages

45

Series

Abstract

Polyphonic pitch recognition has only been sparsely studied in the context of guitar audio, largely due to the complexity of the task. However, recent advances in neural networks have significantly expanded what is possible in this domain. This work introduces a novel approach based on a Fully Convolutional Network, which is both faster and more accurate than the current state of the art. The proposed model outperforms FretNet by 8% in Raw Pitch Accuracy, while integrating the most important features of both PENN and FretNet. In addition, modifications to the widely used Guitarset dataset are proposed, further enhancing the performance of the model. The presented pitch recognition approach has potential applications in automatic music transcription, audio synthesis, and sample-based audio playback.

Description

Supervisor

Välimäki, Vesa

Thesis advisor

Juvela, Lauri

Keywords

deep learning, music information retrieval, convolutional neural networks, polyphony, guitar, transcription (music)

Other note

Citation

Permanent link to this item

https://urn.fi/URN:NBN:fi:aalto-202508196303

Collections

[dipl] Sähkötekniikan korkeakoulu / ELEC