Harmonic Resonance Boost — User Guide

A phase-preserving harmonic magnitude comb that emphasizes frequency bands around integer multiples of a chosen fundamental while attenuating the spectrum between them.

Author: Shai Cohen Affiliation: Department of Music, Bar-Ilan University, Israel Version: 0.6.2 (2026) License: MIT License Repo: https://github.com/ShaiCohen-ops/Praat-plugin_AudioTools

Contents:

What this does Harmonics & resonance Quick start Presets Controls Processing pipeline Channels & level Visualization

What this does

Harmonic Resonance Boost applies a hard-edged frequency-domain gain comb. Frequencies close to integer multiples of a user-defined fundamental f₀ are multiplied by Harmonic boost; frequencies outside those bands are attenuated by one of two gains, separated by Mid freq cutoff. The filter multiplies both real and imaginary FFT components by the same real gain, so it changes magnitude while preserving the source phase at every bin.

Important: the script does not detect the source's fundamental. Fundamental frequency defines the harmonic grid you want to impose. A 200 Hz setting therefore boosts neighborhoods around 200, 400, 600, 800 Hz, and so on, whether or not the input is pitched at 200 Hz.

What are harmonics and harmonic resonance?

A harmonic series is a set of frequencies at integer multiples of a fundamental frequency: f₀, 2f₀, 3f₀… . Many periodic musical tones contain energy near such multiples. This processor uses that arithmetic relationship as a spectral selection grid: energy near the grid is emphasized relative to energy between the harmonics.

This is best understood as a magnitude comb filter, not as physical resonance modelling. No resonating body, decay time, or feedback loop is simulated. The word “resonance” describes the resulting emphasis of a chosen harmonic family.

nearest harmonic n = round(f / f0) if n ≥ 1 and |f − n·f0| < half_width: gain = harmonic_boost else if f < mid_freq_cutoff: gain = low_mid_attenuation else: gain = high_freq_attenuation

If the half-width reaches roughly half the fundamental spacing, adjacent boosted cells meet or overlap. The upper spectrum then becomes continuously boosted rather than alternating between boosted and attenuated bands.

Quick start

Select exactly one Sound.
Run Harmonic_Resonance_Boost.praat.
Choose a preset, or use Custom.
Set the harmonic fundamental and half-width. Wider bands retain more of the original spectrum around each harmonic.
Choose mono or stereo output and click OK.

Presets

Preset	f0	Half-width	Boost	Low/Mid	High	R widening
Sub Drone	50 Hz	20 Hz	4.0×	0.15	0.05	8 Hz
Industrial Hum	60 Hz	15 Hz	5.0×	0.10	0.02	5 Hz
Laser Comb	200 Hz	8 Hz	6.0×	0.05	0.02	3 Hz
Alien Voice	333 Hz	40 Hz	3.5×	0.08	0.03	20 Hz
Metallic Ring	666 Hz	25 Hz	5.5×	0.06	0.04	12 Hz
Glass Bells	1200 Hz	35 Hz	3.0×	0.20	0.15	25 Hz
Celestial Pad	528 Hz	80 Hz	2.0×	0.25	0.15	40 Hz
Swarm	77 Hz	12 Hz	4.5×	0.12	0.06	6 Hz

Presets replace the listed spectral values but leave the output switches, including Create stereo, Scale peak, visualization, and playback, under user control.

Controls

Control	Meaning
Fundamental frequency	Spacing of the harmonic grid. Must be below Nyquist.
Harmonic half-width	Boosted region on each side of every harmonic. This is a half-width: a 20 Hz value means approximately ±20 Hz.
Harmonic boost	Linear gain applied inside harmonic cells. 1 = unity; values above 1 boost; 0 suppresses the harmonic cells.
Mid freq cutoff	Boundary between the two non-harmonic attenuation gains.
Low/mid attenuation	Gain for non-harmonic bins below the cutoff; range 0…1.
High freq attenuation	Gain for non-harmonic bins at or above the cutoff; range 0…1.
Create stereo	Builds a two-channel result with different harmonic band widths.
Right-band widening	Added only to the right-channel half-width. It does not shift the harmonic centres.
Scale peak	Final target peak for every non-silent result.

Processing pipeline

Choose the source channel topology.
Transform the source to a Spectrum.
Apply the exact harmonic gain law directly to the complex Spectrum.
Convert back to Sound and trim FFT padding to the original duration.
For stereo, combine the independently filtered L/R results.
Target-normalize the final non-silent Sound to Scale peak.

Channels, duration and level

Create stereo ON + mono input: one source spectrum feeds both sides; left uses the base half-width and right uses the widened half-width.
Create stereo ON + stereo/multichannel input: source channels 1 and 2 are filtered independently. Channels above 2 are not included in the output.
Create stereo OFF: multichannel input is folded to mono before filtering; output is mono.
Duration and source sample rate are retained.
Scale peak is target normalization, not a clipping-only ceiling. Any non-silent result is scaled to that peak value.

Visualization

The Picture window shows original and filtered waveforms, the original and pre-output-scaling spectra, the exact left/right harmonic gain laws, and a summary of f0, band widths, boost/attenuation and comb coverage. The filter-response plot uses the same nearest-harmonic rule as the audio processing.