Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Analyzing Pitch Content in Traditional Ghanaian Seperewa Songs

Kelvin L Walls, Iran R Roman, Kelsey Van Ert, Colter Harper, Leila Adu-Gilmore

Abstract

This study examines the pitch content in traditional Ghanaian seperewa (Akan harp-lute) songs, utilizing a unique dataset from field recordings of the mid-twentieth century. We selected 71 songs and used Demucs to isolate vocals from instrumental tracks. We then retrieved the F0 content from these isolated tacks and applied Gaussian Mixture Models (GMM) to approximate musical scales. Comparative F0 analysis between vocals and seperewa revealed higher microtonal deviations from equal temperament in vocal tracks. We also note challenges in using MIR tools for musical scale approximation in non-Western music. Our research contributes to the quantitative study of pitch in traditional music of Sub-Saharan Africa.

Analyzing Pitch Content in Traditional Ghanaian Seperewa Songs

Abstract

This study examines the pitch content in traditional Ghanaian seperewa (Akan harp-lute) songs, utilizing a unique dataset from field recordings of the mid-twentieth century. We selected 71 songs and used Demucs to isolate vocals from instrumental tracks. We then retrieved the F0 content from these isolated tacks and applied Gaussian Mixture Models (GMM) to approximate musical scales. Comparative F0 analysis between vocals and seperewa revealed higher microtonal deviations from equal temperament in vocal tracks. We also note challenges in using MIR tools for musical scale approximation in non-Western music. Our research contributes to the quantitative study of pitch in traditional music of Sub-Saharan Africa.

Paper Structure

This paper contains 11 sections, 3 figures, 2 tables.

Table of Contents

  1. Introduction and Motivation
  2. Activating An Archival Ghanaian Music Dataset
  3. Methods
  4. Scale Aproximation Pipeline (SAP).
  5. Results
  6. Finding scales in isolated seperewa & vocal tracks.
  7. Deviation of scales from equal temperament.
  8. Microtonal inflections of scale degrees.
  9. Comparing the seperewa and vocal scales directly.
  10. Discussion
  11. Conclusion

Figures (3)

  • Figure 1: How equal-tempered are the scales of songs? This figure shows the density of the error $\epsilon_S$ (in cents) between the equal-tempered scale and the scale we approximated with our SAP for each song in the corpus. On the x-axis, an $\epsilon_S$ at or close to zero would correspond to an equal-tempered scale, while higher $\epsilon_S$ values denote deviation from equal temperament. Separate histograms show the density of $\epsilon_S$ for the seperewa and the vocals.
  • Figure 2: How microtonally flat or sharp are individual scale degrees in our corpus with respect to equal temperament? The density of all scale components found with our SAP on the entire corpus. The vertical red lines correspond to a twelve-tone equal-tempered scale. The x-axis is the frequency distance from the tonic (in cents). The top plot shows the component density in the seperewa and the bottom is the corresponding plot for vocals.
  • Figure :