Dynamic EQ has been the default fix for resonant problems in mastering chains for over a decade. You set a frequency band, dial in a threshold, and the plugin attenuates whenever the signal crosses it. Simple, predictable, and good enough for most jobs.
But "good enough" starts to fray the moment a problem frequency moves. A vocal overtone that sits at 2.4 kHz on one note and drifts to 2.7 kHz on the next forces you into a compromise. Widen the band, and you dull everything around the problem. Narrow it, and you miss half the occurrences. Additive resynthesis EQ, as implemented in tools like the PARTIAL EQ section of the REISEI mastering processor, takes a fundamentally different approach by locking onto the harmonic itself and following it wherever it goes.
That distinction sounds incremental until you hear what it does to a dense master. You stop fighting the spectrum and start editing the source.
Harmonic-tracking EQ built on additive resynthesis decomposes the incoming signal into individual partials and tracks each one across the spectrum in real time. When a vocalist bends from A3 to C4, the tracked overtone moves with the note. Your gain reduction follows the problem rather than hoping the problem stays inside your bandwidth. The practical result is that you can set a 0.8 dB cut on a single aggressive partial and leave the surrounding air, breath, and articulation completely untouched.
This is closer to harmonic editing than frequency-based correction. You are not carving a hole in the spectrum. You are reaching into the harmonic structure and adjusting one component.
The difference matters most on complex program material. A ring frequency at 4.1 kHz in a mastered guitar track might overlap with cymbal wash and vocal sibilance energy. A dynamic EQ cut at that frequency drags all three down together. Inverting the specific partial removes only the ringing component. The cymbal and the vocal pass through unchanged.
You will not need this on every master. Maybe one track in ten has a problem that specific. But when it does, no amount of dynamic EQ finesse replicates the result.
Additive resynthesis does not work that way. The processor reconstructs the signal from tracked partials, so the output is linear by design. There is no filter curve generating phase rotation in the first place. You are not choosing between minimum-phase coloration and linear-phase transparency. The architecture only produces one kind of result, and it happens to be the transparent one.
For a mastering engineer running a full chain of EQ, compression, and saturation, that guarantee removes one variable from every bypass comparison. You do not have to wonder whether the tonal change you hear is the correction you intended or a phase artifact of the correction. The processor adds no distortion and introduces no phase shift. It simply changes the amplitude of the partials you touched.
But "good enough" starts to fray the moment a problem frequency moves. A vocal overtone that sits at 2.4 kHz on one note and drifts to 2.7 kHz on the next forces you into a compromise. Widen the band, and you dull everything around the problem. Narrow it, and you miss half the occurrences. Additive resynthesis EQ, as implemented in tools like the PARTIAL EQ section of the REISEI mastering processor, takes a fundamentally different approach by locking onto the harmonic itself and following it wherever it goes.
That distinction sounds incremental until you hear what it does to a dense master. You stop fighting the spectrum and start editing the source.
Pitch following changes what your equalizer can actually hear
A dynamic EQ sits in a fixed frequency window and waits. It reacts to amplitude, not to musical content. FabFilter's spectral dynamics mode in Pro-Q 4 narrows the trigger to specific frequencies within that window, which is a genuine improvement, but the window itself still does not move. The processor has no concept of pitch. It cannot tell you that the third harmonic of a bass note is climbing as the player slides up the neck.Harmonic-tracking EQ built on additive resynthesis decomposes the incoming signal into individual partials and tracks each one across the spectrum in real time. When a vocalist bends from A3 to C4, the tracked overtone moves with the note. Your gain reduction follows the problem rather than hoping the problem stays inside your bandwidth. The practical result is that you can set a 0.8 dB cut on a single aggressive partial and leave the surrounding air, breath, and articulation completely untouched.
This is closer to harmonic editing than frequency-based correction. You are not carving a hole in the spectrum. You are reaching into the harmonic structure and adjusting one component.
Anti-phase cancellation removes a partial cleanly
Most top results describing harmonic-tracking EQ mastering workflows stop at "you can lift or reduce a partial." They skip the most interesting operation available. REISEI's PARTIAL EQ can invert a tracked partial so that it cancels in anti-phase against the original signal. That is not the same as pulling down a fader on a frequency band. Band attenuation always leaves residual energy at the edges of the filter slope. Anti-phase cancellation targets the exact waveform component and subtracts it, leaving the fundamental and every neighboring harmonic intact.The difference matters most on complex program material. A ring frequency at 4.1 kHz in a mastered guitar track might overlap with cymbal wash and vocal sibilance energy. A dynamic EQ cut at that frequency drags all three down together. Inverting the specific partial removes only the ringing component. The cymbal and the vocal pass through unchanged.
You will not need this on every master. Maybe one track in ten has a problem that specific. But when it does, no amount of dynamic EQ finesse replicates the result.
Linearity is architectural, not a mode you toggle
Dynamic EQ and spectral dynamics processing introduce phase shifts. FabFilter's own documentation notes that spectral mode forces linear-phase processing on the affected bands, implying that the rest of the signal path does not run linear-phase by default. You toggle a mode to compensate. The correction is bolted on.Additive resynthesis does not work that way. The processor reconstructs the signal from tracked partials, so the output is linear by design. There is no filter curve generating phase rotation in the first place. You are not choosing between minimum-phase coloration and linear-phase transparency. The architecture only produces one kind of result, and it happens to be the transparent one.
For a mastering engineer running a full chain of EQ, compression, and saturation, that guarantee removes one variable from every bypass comparison. You do not have to wonder whether the tonal change you hear is the correction you intended or a phase artifact of the correction. The processor adds no distortion and introduces no phase shift. It simply changes the amplitude of the partials you touched.
