Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

GuitarFlow: Realistic Electric Guitar Synthesis From Tablatures via Flow Matching and Style Transfer

Jackson Loth, Pedro Sarmento, Mark Sandler, Mathieu Barthet

TL;DR

GuitarFlow tackles the challenge of controllable, expressive electric guitar synthesis from tablatures by combining a fast tab-to-audio rendering step with Flow Matching-based style transfer to achieve realism with limited data. The method encodes both synthetic and real guitar DI into a shared latent space and learns a velocity field via Flow Matching to transfer synthetic timbre toward real guitar sounds, enabling efficient training and inference. Evaluations on the GOAT dataset show objective improvements in realism and a listening test confirms perceptual gains, illustrating Flow Matching as a promising approach for data-efficient generative audio. The work highlights practical benefits for guitar synthesis and motivates further exploration of Flow Matching in expressive, instrument-specific generation.

Abstract

Music generation in the audio domain using artificial intelligence (AI) has witnessed steady progress in recent years. However for some instruments, particularly the guitar, controllable instrument synthesis remains limited in expressivity. We introduce GuitarFlow, a model designed specifically for electric guitar synthesis. The generative process is guided using tablatures, an ubiquitous and intuitive guitar-specific symbolic format. The tablature format easily represents guitar-specific playing techniques (e.g. bends, muted strings and legatos), which are more difficult to represent in other common music notation formats such as MIDI. Our model relies on an intermediary step of first rendering the tablature to audio using a simple sample-based virtual instrument, then performing style transfer using Flow Matching in order to transform the virtual instrument audio into more realistic sounding examples. This results in a model that is quick to train and to perform inference, requiring less than 6 hours of training data. We present the results of objective evaluation metrics, together with a listening test, in which we show significant improvement in the realism of the generated guitar audio from tablatures.

GuitarFlow: Realistic Electric Guitar Synthesis From Tablatures via Flow Matching and Style Transfer

TL;DR

GuitarFlow tackles the challenge of controllable, expressive electric guitar synthesis from tablatures by combining a fast tab-to-audio rendering step with Flow Matching-based style transfer to achieve realism with limited data. The method encodes both synthetic and real guitar DI into a shared latent space and learns a velocity field via Flow Matching to transfer synthetic timbre toward real guitar sounds, enabling efficient training and inference. Evaluations on the GOAT dataset show objective improvements in realism and a listening test confirms perceptual gains, illustrating Flow Matching as a promising approach for data-efficient generative audio. The work highlights practical benefits for guitar synthesis and motivates further exploration of Flow Matching in expressive, instrument-specific generation.

Abstract

Music generation in the audio domain using artificial intelligence (AI) has witnessed steady progress in recent years. However for some instruments, particularly the guitar, controllable instrument synthesis remains limited in expressivity. We introduce GuitarFlow, a model designed specifically for electric guitar synthesis. The generative process is guided using tablatures, an ubiquitous and intuitive guitar-specific symbolic format. The tablature format easily represents guitar-specific playing techniques (e.g. bends, muted strings and legatos), which are more difficult to represent in other common music notation formats such as MIDI. Our model relies on an intermediary step of first rendering the tablature to audio using a simple sample-based virtual instrument, then performing style transfer using Flow Matching in order to transform the virtual instrument audio into more realistic sounding examples. This results in a model that is quick to train and to perform inference, requiring less than 6 hours of training data. We present the results of objective evaluation metrics, together with a listening test, in which we show significant improvement in the realism of the generated guitar audio from tablatures.
Paper Structure (20 sections, 2 equations, 3 figures, 1 table)

This paper contains 20 sections, 2 equations, 3 figures, 1 table.

Table of Contents

  1. Introduction
  2. Background
  3. Guitar Tablatures
  4. Style Transfer
  5. Music Instrument Synthesis
  6. Methodology
  7. Flow Matching
  8. Method
  9. Model
  10. Experiments
  11. Data and Training
  12. Evaluation
  13. Results and Discussion
  14. Objective Metrics
  15. Listening Test
  16. ...and 5 more sections

Figures (3)

  • Figure 1: Example of a guitar tablature, from the Guitar Pro editing software.
  • Figure 2: Training and inference using GuitarFlow.
  • Figure 3: Boxpot with mean indicators for the MOS results of both the DI and amplifier conditions of the listening test. Mean scores in each group marked by white dots.