Publication     Conference     Thesis

Style-Specific Harmonization

Motivation
  Automated music composition aims at applying computer algorithms to create music. It enables musical novices with limited knowledge to create new music and helps professional musicians to enhance their creative process. More importantly,
it fosters collaboration between musicians and non-musicians, leading to diverse applications like music for videos, games, and therapy.
   
Research Goal
  Our research goal is to develop an algorithm called FlexChord for style-specific generation of chords to support a melody.

FlexChord generates chords according to the chord progression of a user-specified reference music that conveys the desired style. It uses direct optimization that does not require training. Therefore, it is very flexible and efficient.

For comparison:

Comparison with DNNs
  original: original human-composed chords
VTHarm: DNN based on transformer
DAT-VAE: DNN based on variational autoencoder

 
Melody: Colors of the Wind (pop)
melody
original

VTHarm
  dissonant chord
DAT-VAE
  dissonant chord
FlexChord
   
Melody: Black Orpheus (pop)
melody
original
VTHarm
  dissonant chord
DAT-VAE
  dissonant chord
FlexChord
   
Custom Styles
  FlexChord generates chord progressions according to specific reference styles.

 
Jazz Style
Melody: Israel (jazz)
melody
original

  jazz-style chord, heightened tension
FlexChord

  jazz-style chord, heightened tension
   
Various Styles
Melody: Colors of the Wind
melody
pop
jazz
  jazz-style chord, heightened tension
Greek
  Greek-style chord, slightly unresolved tension
   
Spooky Style
Melody: In the Hall of the Mountain King
Reference: Psycho
melody
original
FlexChord
   
Emotion Styles
Melody: Ode to Joy
Joyous
Sad

Researchers
>> Ms. Stephanie Y. H. Lew, Ph.D. student, Dept. of Computer Science, Natioanl University of Singapore
>> A/Prof. Leow Wee Kheng, Dept. of Computer Science, Natioanl University of Singapore
>> A/Prof. Kat R. Agres, Centre of Music and Health, National University of Singapore
   
Publications
>> S. Y. H. Lew, W. K. Leow, K. R. Agres. Style-Specific Harmonization by Multi-Objective Optimization without Training, International Journal of Music Science, Technology and Art, 7(2): 58-73, 2025.
>> S. Y. H. Lew, W. K. Leow, K. Agres. Style-Specific Harmonization by Multi-Objective Optimization without Training, In Proc. Int. Conf. on New Music Concepts, March 2025.

5 December 2025