Spectral Compressor

Parameters

Threshold — Per-band level (in dB) where compression starts. Unlike the time-domain Compressor (which acts on the full-range signal), this node splits the signal into frequency bands and compresses each one independently — so the threshold applies per band. A loud snare hit only triggers compression in the snare’s frequency range; the bass and high-end keep their dynamics intact. Range goes lower (−92 dB) than the TD compressor because it’s reading per-band magnitudes from the spectral data.

Ratio — How aggressively each band’s signal above its threshold is reduced. Standard compression ratios apply (4:1 = gentle leveling, 20:1 = near-limiting), but the effect is per-band — high ratios on a spectral compressor produce a “spectral evening out” character rather than the obvious pumping of a TD compressor. Useful for de-essing-style work, taming resonances, and overall spectral balance.

Attack — How fast each band’s compression engages once its level crosses threshold, in ms. Fast attack catches transients in each band; slow attack lets the per-band attack-edge through. Because compression is per-band, fast attacks here don’t dull the overall transient like a wide-band fast-attack would — only the offending band’s transient is shaped.

Release — How fast each band releases its gain reduction, in ms. Slow release in spectral compression sounds different than TD release: instead of a single GR envelope chasing the program, you have many per-band envelopes, which can produce a smooth, breathing “spectral fog” if release is too long. Start with 50–150 ms and adjust by ear.

Makeup — Post-compression makeup gain in dB applied to the full output. Compensates for the average level reduction across all bands. Less critical than on a TD compressor since per-band compression often only reduces a few specific bands rather than the entire signal.

Additional controls

Bands — Number of frequency bands the spectrum is split into for independent compression, 1 to 32. More bands = finer resolution and more transparent results (closer to “everywhere there’s a peak, attenuate it”). Fewer bands behave more like a multi-band compressor with larger ranges. 8–16 is a balanced default; 32 for surgical work; 4–8 for clearly audible multi-band character.

Detection Mode — How each band’s level is measured. Peak — instantaneous per-bin magnitude. RMS — short-window average per bin. RMS is generally more musical for spectral work; Peak is more reactive and can sound chattery on tonal material.

Knee — Soft-knee transition width in dB applied per-band. Same role as on the TD compressor — softens the engagement of compression around the threshold for a more transparent feel. 0 = hard knee, 6+ dB = smooth/transparent.

About Spectral Compressor

Spectral dynamics nodes are unique to K2K’s domain: they operate on the spectral data directly, frame-by-frame, with independent control over each frequency band. A spectral compressor doesn’t just compress “the loud parts” — it compresses which frequencies are loud at each moment, leaving everything else untouched. This produces results impossible with full-range time-domain compression: cleaning up build-ups in mid frequencies without dulling the whole mix, taming a single resonant note in a chord, or sitting transients perfectly without losing punch in adjacent bands. Use the TD compressor for global glue and pumping character; use the Spectral Compressor for surgical balance.

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