Spectral Roughness

Parameters

Roughness — Strength of the per-bin variation, 0–100%. 0 = clean pass-through. Low values (10–30%) add subtle micro-variation to bin magnitudes — the kind of “alive” feel that distinguishes analog from digital. Mid (40–60%) adds clearly audible noise/movement — tape-like character. High (80–100%) approaches dirty, broken-radio territory.

Rate — How fast the per-bin variation evolves over time, in Hz, 0.1–20. Slow rates (0.1–1 Hz) produce slow drifts, gradual warmth-evolution. Mid (2–8 Hz) feels like organic movement — vinyl wow/flutter, breathing tape. Fast (10–20 Hz) becomes tremolo-like or shaky.

Correlation — How correlated the per-bin variations are with each other, 0–1. 0 = each bin varies completely independently (pure noise; produces a hissy, granulated character). 1 = all bins vary together (the whole spectrum modulates as one — closer to a tremolo). 0.5–0.8 produces the most natural-feeling “analog warmth” where neighboring bins move together but distant bins drift independently.

Frequency Bias — Where in the spectrum the roughness is most active, −1 to +1. 0 = uniform (all frequencies equally affected). Negative values bias toward low frequencies (roughens bass and lower mids more — useful for “fat warm bass” character). Positive values bias toward highs (roughens treble more — useful for “tape hiss” character on the top end).

Mix — Equal-power blend between dry (0%) and the roughened signal (100%).

Output Gain — Output level in dB, −24 to +24.

Additional controls

Type — Noise generator that drives the variation:

• Smooth — slowly-varying smooth random noise; subtle, organic drift.
• Filtered — bandpass-filtered noise; more focused per-bin movement.
• S&H — sample-and-hold; stepped/jagged variation, more chunky character.
• Gauss — Gaussian-distributed random; statistically natural variation, similar to Smooth but with more “spike” potential at extremes.

For analog warmth simulation, Smooth or Gauss work best. For more obvious character or sound-design use, try S&H for stepped roughness or Filtered for focused per-bin grit.

About Spectral Roughness

Digital signals are mathematically smooth — every bin is exactly what the algorithm calculated. Real-world (analog) signals have constant micro-variations: tape flutter, electrical noise, capacitor drift, mechanical resonance. Spectral Roughness adds those variations back into the digital signal at the bin level. The result is “warmth” or “movement” or “life” that sterile digital sources often lack. Use it to: warm up clean synths, add tape-like character to recordings, simulate the small imperfections of analog gear, or as a creative texture-shaping tool. The type, rate, and correlation controls let you shape the character of the variation; frequency_bias lets you target specific spectral regions.

Generated 2026-05-05 from K2K_Dev@96730bdc by scripts/gen_lexique.py. Edit _intros/ or _overrides/, not this file.