Source-linked AI summary
Klangfarbenakkord and Klangfarbenharmonien Metric Space Models for Music on Informational Geometry 1
Yusei Tamura, Shigekazu Ishihara, Ken Ito
TL;DR
Western music theory largely represents sound through pitch rather than timbre, leaving spectral characteristics under-theorized. The paper introduces geometric harmony using Wasserstein-based relations among instrumental timbres and shows applications to timbre separation and compatible fingering design in ensemble playing.
Problem
Conventional music theory represents sound primarily through pitch rather than timbre, leaving spectral characteristics insufficiently addressed.
Method
The paper models instrumental timbres as spectral elements, applies Wasserstein distances and deviations, and preserves these distances through Persistent Homology mapping.
Results
The framework separates clusters of distinctive timbres and supports a new fingering method with notably high compatibility.
Takeaways & Limitations
The framework provides a basis for devising fingering methods intended to improve ensemble quality.
Takeaways & Limitations
The paper identifies extending the ensemble treatment to n parts (n≥4) as a limitation.
Abstract
from arXiv · showhide
This paper deals with the introduction of "geometric harmony", a discipline that explicitly addresses the spectral characteristics of musical gamut. The framework of Western music, from Renaissance to the present, represents sound in terms of "pitch"-as is evident from its five-line staff notation system-and employs the fundamental frequency as its representative value, 440 Hz, etc. In this paper, by taking the timbres of specific individual instruments as elements and examining the Wasserstein distance between two voices, and Wasserstein deviations between three or more voices, we demonstrate that it is possible to expand the system whilst retaining the entire framework of conventional music theory. At the same time, as an example of practical utility in ensemble playing, we provide a detailed account of the two-voices affinity of the "Throat G" on clarinet, a note known for its fragility in ensemble contexts.
General Overview
The section frames a longstanding theoretical question about treating timbre as musically fundamental rather than reducing music to pitch. It positions the paper as a 2020s theoretical and practical response developed from composers’ and players’ perspectives.
- Conceptual foundation: The section presents timbre as the broader realm in which pitch is one dimension, challenging the conventional distinction between timbre and pitch.This conceptual position is attributed to Schönberg’s discussion of timbre and pitch.
- Problem statement: The unresolved status of a theory of timbre is underscored by the question, “Who would dare to demand a theory here?”The question follows the discussion of ‘Klangfarbenmelodie’ and its demand for highly refined perception.
- Historical context: Schönberg’s ‘dodecaphonic technique’ addressed musical organization through pitch sequences, but did not answer the broader question concerning timbre.The section presents ‘Klangfarbenmelodie’ as melody of timbres and distinguishes this from dodecaphony’s focus on pitch.
- Motivation: The paper expands Schönberg’s question about timbre theoretically from composers’ and players’ standpoints, offering a 2020s-level answer with concrete musical examples.The stated aim is to provide an answer based on contemporary human understanding and demonstrate its relevance to music.
１ Stochastic Spectra and transport problem
The section models perceived pitch as stochastic spectral events and represents instrumental timbres through normalized short-time Fourier spectra. It then uses one-dimensional Wasserstein transport in frequency space to quantify timbral change as a scalar in Hz, including spectral-shape changes beyond center-of-mass shifts.
- Stochastic spectra: The framework emphasizes that pitch is only a limited attribute of complex, time-varying sound, illustrated by Edge perception in band noise.Pitches near bandpass edges may be perceived while the region between the Edges remains unperceived.
- Stochastic spectra: Normalized short-time Fourier spectra are treated as probability density functions for listeners’ perceived frequencies, while perceived pitch is modeled as Poisson-distributed stochastic events.This approach replaces direct cochlear frequency processing with short-time Fourier analysis of instrumental timbre or speech signals.
- Transport problem: One-dimensional Wasserstein distance compares spectra by optimally transporting frequency distributions, with distance measured in Hz.The method is applied to transfer the chest-voice spectrum toward the yodel or falsetto spectrum and defines the transport dimension in frequency space.
- Transport problem: Wasserstein distance is at least as large as the spectral center-of-mass shift, capturing additional timbral change caused by altering spectral shape [5].The resulting change is expressed as a scalar frequency value in Hz.
- Transport problem: The proposed frequency-based timbre-change estimate is presented as convenient and useful for practical music and musicians, providing a previously unavailable representation.The section motivates the measure as a way to estimate timbral differences directly in frequency units.
two spectra
Wasserstein self-distance matrices reveal instrument-specific spectral structures: flute register shifts contrast with viola string-related changes. The framework also exposes clarinet Throat G divergence and supports symmetric cross-instrument matrices for analyzing timbral relationships.
- Flute spectra show a distinct register shift where the harmonic series changes, whereas viola spectra show finer similarity changes associated with string changes.Dark blue denotes low Wasserstein distance, while green and yellow denote higher distance and lower spectral similarity.
- A direct-product construction combines flute and viola timbres into an expanded mutual distance matrix whose diagonal blocks are self-distances and off-diagonal blocks are mutual distances.The arrangement produces a symmetric matrix, although direct expansion is unnecessary when examining only mutual Wasserstein distances.
- The symmetric expanded matrix could support a principal-component-analysis-style treatment using diagonalization, eigenvalues, and eigenvectors.This possible analysis is identified but not developed in the paper.
- The clarinet’s Throat G region, extending approximately from F to A, has spectrally divergent timbres that blend poorly despite being physically soft.Nearly all finger holes are open at G, making the register difficult to control and challenging for harmonic unity with other instruments.
- The clarinet’s large Wasserstein distances around Throat G can guide fingering choices that preserve solo or soloistic lines against accompaniment.Spectral divergence is therefore not inherently disadvantageous when physically soft notes need to remain audible.
3. Topological mapping and examples of its application in ensemble
Persistent Homology preserves Wasserstein distances while mapping clarinet timbres into clusters, separating distinctive registers and the clarinet’s “Throat G”. Wasserstein comparisons then identify compatible unison fingerings and support objective ensemble-oriented fingering design.
- 3. Topological mapping and examples of its application in ensemble: Persistent Homology preserves Wasserstein distance while mapping full-range B♭ clarinet timbres into clusters that separate distinctive registers and isolate the “Throat G”.The mapping extracts and visualizes connections between multidimensional data points across scales.
- 3. Topological mapping and examples of its application in ensemble: In Beethoven’s Symphony No. 8, the clarinet may use unprotected “Throat G” because of phrase dynamics, requiring other instruments to compensate for ensemble quality.Alternative fingerings exist, but the unprotected fingering is often adopted in this passage.
- 3. Topological mapping and examples of its application in ensemble: Wasserstein distance identified a clarinet “Throat G”–flute standard-fingering combination as having the smallest spectral-shape difference in unison.The difference was not discernible by simply listening to isolated notes.
- 3. Topological mapping and examples of its application in ensemble: Thirteen newly devised oboe F fingerings were evaluated against clarinet “Throat G”, establishing a fingering method with notably high compatibility.The Wasserstein distance matrix provided the basis for devising fingering methods intended to improve ensemble quality.
- 3. Topological mapping and examples of its application in ensemble: Professor TURNOVSKY’s bassoon fingering produced a small Wasserstein distance regardless of whether standard or alternative clarinet “Throat G” fingering was used.The technique can be unstable because it uses an old cross-fingering, yet its effectiveness was objectively measured.
Klangfarbenharmonie
The section extends Wasserstein-based timbre affinity from two voices to three-voice Klangfarbenakkorde, using triangle area and Wasserstein Timbre Deviation to quantify harmonic unity or dispersion. These measures enable serialization of chord voicings and timbral progressions within a quantified model of Klangfarbenharmonie.
- Klangfarbenharmonie: Three-voice timbral harmony is quantified by a Wasserstein Triangle whose area measures Klangfarbenakkord affinity or dissociation in [Hz²].The triangle is constructed from the three pairwise Wasserstein distances, which satisfy the triangle inequality.
- Klangfarbenharmonie: Because different close and open voicings share harmonic function but differ in timbral dispersion, the metrics can serialize instrumental Klangfarbenakkord arrangements by affinity or dissociation.The example evaluates pairwise distances from each instrument’s spectrum at its respective pitch, including Cor Anglais, bass clarinet, and bassoon configurations.
- Klangfarbenharmonie: The triangle-area construction does not generally extend to n ≥ 4 voices, because arbitrary Wasserstein edge lengths may not form valid higher-dimensional simplices.Quartet and woodwind-quintet applications are therefore outside this paper, although similar methods may be possible in future work.
- Klangfarbenharmonie: Wasserstein Timbre Deviation σ = √M_2 provides an alternative [Hz]-scaled measure of triadic timbral dispersion and spectral idiosyncrasy.For the same chord configurations, deviation can differ from triangle area while still evaluating timbral affinity.
- Klangfarbenharmonie: This serialization supports quantified timbral progressions that can sustain unceasing timbral tension or realization, independently of traditional cadence-based tension and release.The resulting model quantifies individual timbral chords within a Klangfarbenharmonie framework based on Schönberg’s original idea.
5. From “die Klangfarbe” to “die Sprechstimme”
This section traces timbre and speech research from Helmholtz and early synthesis methods to machine-learning-based timbre geometrization. It presents an information-geometric framework that combines sound series, timbre series, and spoken-language timbral variation into one system.
- From “die Klangfarbe” to “die Sprechstimme”: Earlier work progressed from Helmholtz’s physiological acoustics and vowel synthesis to vocoder-based speech analysis and Fandt’s source-filter synthesis.Fandt’s approach produced vowel synthesis in OVE1 (1953) and speech synthesis including consonants in OVE2 (1962).
- From “die Klangfarbe” to “die Sprechstimme”: The study differs from prior timbre mapping by avoiding representative instrumental “lattice points” and instead computing optimal-transport distances between multiple timbre spectra.This aims to make quantities associated with ensemble playing computable, advancing toward Schoenberg’s classic question.
- From “die Klangfarbe” to “die Sprechstimme”: Information Geometry proposed by Shunichi AMARI reframes Schoenberg’s three themes as a single system of problems.The framework treats speech and music spectra as probability density functions and derives spectral series for timbre, speech, and their coupling.
- From “die Klangfarbe” to “die Sprechstimme”: The resulting general-purpose information geometry for music combines sound series, timbre series, and spoken language as dynamical variation in timbre.These components form a triadic information-geometric structure expressed through a mathematical product.
Klangfarbenharmonielehre · Notes and references
The paper argues that as Wasserstein metrics between timbres approach zero, classical music theory remains intact, extending the framework beyond any specific genre. This supports applications ranging from popular and traditional music to cultural rituals, speech, and sound.
- Klangfarbenharmonielehre: The concluding analogy invokes Paul Ehrenfest’s 1927 result that microscopic quantum-mechanical expectation values recover Newtonian mechanics in the macroscopic limit.This physical analogy motivates interpreting vanishing timbral differences as a limit that preserves classical musical structures.
- Klangfarbenharmonielehre: As Wasserstein metrics asymptotically approach zero and timbral differences degenerate, classical music theory, including harmony, remains intact.The paper frames this as analogous to the Ehrenfest Theorem, whose quantum-mechanical expectation values agree with Newtonian mechanics in the macroscopic limit.
- Klangfarbenharmonielehre: The preserved classical framework indicates that the approach is not limited to a specific musical genre.The paper presents genre-independence as the basis for broader musical application.
- Klangfarbenharmonielehre: The framework could apply across commercial pop music, traditional ethnic music, cultural rituals, speech, and other sound-related contexts.These examples are presented as possible application domains rather than demonstrated empirical evaluations.
- Klangfarbenharmonielehre: The paper characterizes genre-independent musical utility as a central practical consequence of the approach.This utility is stated to hold regardless of genre.
- Klangfarbenharmonielehre: The paper connects this broad musical utility to Max Mathews’s early emphasis on computer applications of music.The supplied passage identifies Mathews as a related reference but does not specify a particular work or bibliography marker.