home/ atoms/ spectral-analysis-for-scale-design

Designing a scale for an inharmonic instrument requires FFT analysis followed by computing dissonance curve minima

To find the natural scale for an inharmonic instrument: (1) record the instrument and apply FFT analysis (using windowing to reduce spectral leakage); (2) identify the most significant spectral peaks (simplify: keep only the peaks that survive across multiple recordings and analysis windows); (3) construct the simplified spectrum as a list of partial frequencies and amplitudes; (4) compute the dissonance curve by sweeping the interval ratio from 1:1 to 2:1 (or beyond), at each interval calculating the total sensory dissonance between all pairs of partials of the two sounds; (5) identify local minima — these are the consonant intervals; (6) the set of minima defines the related scale. Practical considerations: spurious peaks cause extra scale steps; missing peaks cause missing steps; analyzing multiple recordings improves reliability.

Examples

Tingshaw bell: FFT reveals peaks at 2368, 4784, 5553, 7921, 10103 Hz. Simplified spectrum with those five peaks. Compute dissonance curve → minima at specific ratios → these define a pentatonic-like scale for the tingshaw. Chaco rock: multiple FFT snapshots averaged to find consistent peaks before scale design.

Assessment

Describe step by step how you would determine what scale to use when composing for a set of metal mixing bowls. At which step is it most critical to simplify the spectrum, and what goes wrong if you include too many spectral peaks or too few?

“important to try and select only the most significant peaks, without missing any, because spurious peaks may cause extra minima in the dissonance curve, whereas missing peaks may cause missing scale steps.”
corpus · tuning-timbre-spectrum-scale-william-a-sethares · chunk 46