Zero-latency peaking filters: what RootEQ does to phase, and what it does not

RootEQ reports zero samples of plugin latency. That is true. Each band is a biquad peaking filter — the same IIR object as a huge number of mixing EQs. IIR bells do not look ahead. They also do not have a linear phase response. Boosts and cuts rotate phase around the center frequency. The audible result is often described as “punchier” or “more analog” than a linear-phase EQ, and sometimes as “smeary” if you stack steep, high-Q bands at low frequencies.

What you are choosing

Minimum-phase / IIR (RootEQ, most stock EQs, most analog emulations): no look-ahead delay. Fine for tracking, live, and most mix work. Phase rotation is part of the sound, especially on steep low-end cuts you are not doing in RootEQ, because RootEQ has no dedicated high-pass — but a low peaking cut still has a phase signature.

Linear-phase / FIR: the magnitude curve you drew, with delay in exchange for less phase distortion. Useful in mastering, parallel processing, and crossovers. Not what RootEQ is.

If someone promises an EQ that is simultaneously steep, linear-phase, and zero-latency, they are selling physics a problem. RootEQ does not make that claim.

Practical consequence

On a bass root at 50 Hz with a narrow Q, you will get the usual IIR pre-ringing-is-not-the-issue, post-resonance-is behavior of a peaking filter. Keep Q moderate unless you are notching a whistle. On vocals and synths in the mids, the phase rotation of a few bells is rarely the thing that ruins a mix. Overlapping eight narrow boosts at the bottom of the piano can be. The 8-root cap is a workflow limit and a taste limit.

Shape at extreme negative or high values does not change the filter type. It scales gain. A 200% Shape on a +6 dB lattice is a large boost, with the corresponding phase. Use your ears and the output gain (±24 dB) to keep the plugin from becoming a makeup circus.