Wah-Wah Effect — User Guide

What this does

Wah-Wah Effect creates a moving spectral emphasis by passing each short-time frame through a frequency-domain wah response whose center frequency changes over time. The center can follow a periodic sine LFO or the source amplitude envelope. A broad Hann-like pass band defines the audible region around the moving center, while a Gaussian peak adds controllable resonance at the center frequency.

The processor runs continuously in a short-time Fourier transform (STFT) overlap-add engine rather than cutting the source into hard segments. Analysis and synthesis both use a square-root Hann window, and overlapping frames are divided by an accumulated weight buffer during reconstruction. This produces a smoothly changing filter trajectory while preserving the original duration and sample timing.

Processing pipeline

1. Prepare the source. The selected Sound is copied and shifted internally to start at 0 for processing. Its original start time is restored at the end.
2. Choose output channel geometry. Mono input becomes stereo; input with two or more channels preserves its channel count.
3. Build the modulation trajectory. Sine LFO computes a periodic center-frequency path. Envelope Follower measures frame RMS, converts it to a 0–1 control relative to the file maximum, and smooths it with separate attack and release constants.
4. Analyze overlapping frames. The window is approximately 40 ms and the hop is one quarter of the window, giving 75% overlap.
5. Apply the spectral wah. Each frame is transformed with To Spectrum: "no", multiplied by the moving Hann-like pass response and center resonance peak, and converted back to Sound.
6. Overlap-add reconstruction. A square-root Hann synthesis window is applied and frames are summed at their original sample positions. The accumulated Hann weight is then divided out.
7. Mix dry and wet. The reconstructed wah signal is blended with the original according to Wet_dry_percent.
8. Apply output safety. If enabled, the final signal is attenuated only when its peak exceeds Safety_peak.

Quick start

1. Select exactly one Sound object in Praat.
2. Run Wah-Wah_Effect.praat.
3. Choose a preset, or select Custom.
4. Choose Sine LFO for periodic wah movement or Envelope Follower for source-responsive movement.
5. Set the frequency range, bandwidth, resonance peak, wet/dry amount, and—when using the LFO—stereo offset.
6. Run the script. The result is created as <source>_wahwah.

Modulation modes

Sine LFO

Wah_rate_Hz is the actual LFO rate in cycles per second. For multichannel signals, the phase offset is distributed evenly from 0 cycles on the first channel to Stereo_offset_cycles on the last channel. For mono input, the generated stereo output therefore receives two phase-related wah trajectories.

Envelope Follower

For every STFT frame, the script measures RMS from a mono analysis copy. The RMS values are expressed in dB relative to the loudest frame in the file. Envelope_range_dB defines the level span mapped into the 0–1 control range; values below that span clamp to 0 and the loudest frame maps to 1.

The raw control is smoothed with independent attack and release time constants. That smoothed value then maps linearly between Min_frequency_Hz and Max_frequency_Hz. All output channels use the same envelope-derived center trajectory; the original channel signals themselves remain independent.

Wah response

At each frame, the pass region extends approximately half the selected bandwidth below and above the current center frequency, clipped to 0 Hz and Nyquist when necessary. Each outer edge receives a cosine transition whose width is:

Inside that pass response, a Gaussian peak is centered at the current wah frequency. The linear gain at the exact center is therefore approximately Resonance_peak, while a value of 1 leaves the pass-band center flat.

Presets

Presets override the modulation and wah parameters shown above. Wet_dry_percent, Safety_peak, Draw_response, and Play_result remain user-controlled. LFO presets do not need the envelope attack/release/range fields because those values are ignored in LFO mode.

Parameters

Input & output behavior

Visualization

When Draw_response is enabled, v0.5 draws the AudioTools suite-standard page with:

• input waveform;
• output waveform;
• the center-frequency trajectory over time;
• for Sine LFO, the first-channel and last-channel trajectories so the stereo phase offset is visible;
• for Envelope Follower, the common envelope-derived trajectory;
• a representative midpoint spectral response showing pass-band shape and resonance peak;
• a summary strip with range, bandwidth, resonance, wet percentage, modulation mode, channel behavior, STFT geometry, and final peak.

The visualization uses temporary mono copies only for drawing waveforms. It does not modify the rendered output. In testing, Draw ON and Draw OFF produced sample-identical audio.

Notes & limitations

• Spectral wah, not circuit emulation. The response is constructed directly in each frame spectrum; it does not model a specific analog wah pedal, inductor, transistor ladder, or pedal potentiometer.
• Finite STFT resolution. The 40 ms analysis window determines the frequency-bin spacing. Very narrow requested bandwidths are therefore limited by the frame resolution.
• Time-varying spectral filtering. The response changes once per STFT frame and is reconstructed through overlap-add. It is smooth, but not a sample-by-sample analog filter coefficient update.
• Frequency bounds are protected. The script swaps reversed min/max values and limits the upper center range below Nyquist. Bandwidth and resonance are also constrained to safe internal ranges.
• Safety is a ceiling, not loudness matching. A quieter result remains quieter; the processor never raises gain merely to reach Safety_peak.