A Graph Engine for Guitar Chord-Tone Soloing Education

2025

Read A Graph Engine for Guitar Chord-Tone Soloing Education (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2510-19666.txt.

Introduction

Matthew Keating and Michael Casey built a graph-based tool that helps guitar students practice chord-tone soloing, the jazz skill of improvising using only the notes of the chord currently playing. The practice is a foundation of advanced jazz guitar but is hard to learn alone. Their engine first generates chord-tone arpeggios for each chord in a progression.

Each arpeggio becomes a node in a weighted graph, and edges between consecutive chords are weighted by how smoothly the lines connect through transition tones. The shortest path through the graph is then rebuilt into a complete soloing line. A user-friendly interface lets students enter a progression and receive practice material, turning an abstract theory exercise into something playable.

Catalog record

YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE19 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2510-19666

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

A Graph Engine for Guitar Chord-Tone Soloing Education Abstract We present a graph-based engine for computing chord tone soloing suggestions for guitar students. Chord tone soloing is a fundamental practice for improvising over a chord progression, where the instrumentalist uses only the notes contained in the current chord. This practice is a building block for all advanced jazz guitar theory but is difficult to learn and practice. First, we discuss methods for generating chord-tone arpeggios. Next, we construct a weighted graph where each node represents a chord-tone arpeggio for a chord in the progression. Then, we calculate the edge weight between each consecutive chord’s nodes in terms of optimal transition tones. We then find the shortest path through this graph and reconstruct a chord-tone soloing line. Finally, we discuss a user-friendly system to handle input and output to this engine for guitar students to practice chord tone soloing. 1 Introduction Improvised soloing is a core part of playing jazz music. Traditional jazz bebop standards have a set chord progression the band plays while the soloist improvises musical lines over the chords. Part of learning to improvise on a jazz chord progression is to use chord-tone arpeggios that fit each chord, a process known as chord-tone soloing. A chord tone is defined as a note in a chord. For example, the chord Amin7 comprises the notes A, C, E, and G, so these four notes are all chord tones for Amin7. Learning chord tone arpeggios is particularly difficult on guitar because the guitar is a transposition instrument and has multiple correct ways of playing any one note. The standard guitar has six strings and approximately 22 frets. Without using extended techniques such as bends, each fretted note on the guitar plays a note on the 12-tone chromatic scale. Therefore, the guitar has about 132 possible fretted positions and 11 different positions for each note in the 12-tone scale. A chord tone arpeggio for a four-note chord, standard for jazz chord voicings, is four notes and one-third of the 12-tone chromatic scale. So, approximately 44 positions on the guitar neck play a chord tone for any four-note chord. Chord tone soloing becomes more complicated when expanded to a whole chord progression. The soloist must transition to playing the next chord’s tones when the current chord changes. Generally, the best practice for moving between chord-tone arpeggios minimizes the distance traveled between the transition notes. Therefore, a guitarist must know the chord tone arpeggios in all positions on the guitar fretboard to select the chord tone positions with the best transition notes. Learning these chord tone techniques on jazz guitar is a huge task. Some online resources exist such as textbooks and instructional videos, but many of these resources are behind paywalls. Furthermore, these resources are static content for a specific learning case, such as soloing over a well-known chord progression, and do not extend to every chord progression a student may encounter. In this paper, we detail a system that can generate chord-tone soloing examples for any chord progression. Generating educational material for guitar instruction is a growing topic in computer music. Keating et al. used long short-term memory models to generate guitar chord voicing examples for jazz chord progressions [ 1 ] . We expand on this model by offering an approach for chord tone solos and constraining the model to always produce a meaningful result by using simple graph traversals. Our weighted graph-based engine is similar to a Hidden Markov Model (HMM) approach, common for tablature generation models such as Hori et al. which generates tablature from staff notation [ 2 ] and in piano fingering estimation models such as Nakamura et al. [ 3 ] . HMMs use similar state representations, which we use to represent different chord tone arpeggios, but have probabilities to represent state change and are data-driven. In contrast, our model has linear weights and is weighted with certain quantifiable metrics. Other related research includes improvisation generation models for creating music for performance. Frieler et al. created a data-driven improvisation generation model and tested performance with human listeners [ 4 ] . Pachet’s Continuator uses a Markov model to generate plausible notes for improvisation [ 5 ] . These models have similar goals to our model insofar as they generate improvised solos, but they are focused on machine-generated music for performance rather than music education, and their output is not mapped onto a guitar. Further related work includes writing figured bass lines with rule-based algorithms [ 6 , 7 ] . Our work employs a similar strategy to codify known musical rules for line generation. 2 Graph Engine Our engine for generating guitar chord tone suggestions relies on a weighted graph construction and traversal. 1 1 1 Engine code: https://github.com/mbkeating/chord-tone-generator Each node in this graph represents a set of positions on the guitar that play a chord’s tones. Each edge describes the movement between chords in a progression. We query this graph by traversing the shortest path and assembling the sets of notes in an optimized order. The resulting list of notes is a single-note line to play on guitar which follows the given chord progression. 2.1 Generating guitar chord tones Guitar chord tones are generated by starting at the chord’s root note in some fretboard position and playing each interval in the chord type. For example, an Amin7 chord has the tones A, C, E, and G. A guitarist can start on the fifth fret of the low E string and play the A, then play the C on the eighth fret of the low E string, the E on the seventh fret of the A string, and the G on the fifth fret of the D string. Alternatively, the guitarist can start the arpeggio on a different root note position, giving an equally valid set of chord tones. We could use a set of hard-coded arpeggios to give the model an explicit set of chord-tone arpeggios. The standard arpeggios for most chord shapes are documented in jazz guitar textbooks and online articles. These chord-tone arpeggios can be recorded, indexed by their chord name, and queried when constructing a graph. We can also use a rules-based algorithm to generate these arpeggios. For example, the one used in our preliminary study took a preferred fret stretch constant D , where the user does not want to stretch their fingers more than D frets to reach the arpeggio. We precomputed a 2D array of all notes on a six-string 22-fret guitar G[][] where G[0][0]=E for the low E. Then, given each chord root note and type (where a type implies the intervals/thirds stacking ex. min7 = [3, 4, 3] or minor 3, major 3, minor 3) we iterated over all root note positions and recursively found arpeggios by moving right on the fretboard (no more than D away from the root position) or up the strings. This approach lacks the expert guitarist knowledge that hard-coded arpeggios encode. However, we chose this approach for our preliminary study because the generation works well enough to demonstrate the engine and does not require an expert-curated dataset. 2.2 Constructing the chord tone graph Our graph is formed with two types of objects:

Read the complete text