Vocal Chain
Manual · version 0.4
One window for a lead vocal: high-pass, de-esser, two-band EQ, compressor, shimmer, reverb, tempo-synced delay and width, in that order, with FX ducking, loudness metering and a level-matched bypass.
The idea
Each module is a column with a fader and a four-way selector: Off plus three characters. The character decides what kind of processing the module does; the fader decides how much. That is the whole interface. There are no frequencies, ratios or times to set; those live inside the characters.
Fader convention, the same in every column: 0 is the default amount, +10 the maximum, -10 none. Bass and Treble are the exception: they are ±8 dB with 0 as bypass. Shimmer is bipolar: positive adds air, negative tames it, 0 is bypass.
Signal flow
Input → Balance → High-pass → De-ess → EQ → Compress → Shimmer → dry plus (Reverb, Delay, Width) → Output → loudness meter. The three effects run in parallel and are added to the dry signal; the FX duck pulls their sum down while the vocal is present.
Columns
| Control | What it does |
|---|---|
| Input | Sensitivity, ±10 dB, with a peak meter. Balance ±6 dB under it. High-pass row: Off, 60, 90 or 130 Hz, 12 dB per octave. |
| De-ess | Characters set the band: Soft above 5 kHz, Mid above 6.5 kHz, High above 8 kHz. The fader shifts the threshold; +10 is much more de-essing. The detector compares the band with the whole signal, so it reacts to an ess standing out, not to absolute level, and leaves bright vowels alone. Gain-reduction bar under the fader. |
| Bass | ±8 dB. Sub: shelf at 70 Hz. Lower: bell at 160 Hz. Upper: bell at 320 Hz. |
| Treble | ±8 dB. Bite: bell at 3.2 kHz. Top: shelf at 8 kHz. Roof: shelf at 13 kHz. |
| Compress | Push: 3:1, slow, soft knee. Spank: 6:1, fast, firm. Wall: 20:1 with a soft clip, near a limiter. The fader shifts the threshold: 0 is the character's default, +10 heavy, -10 almost none. Gain-reduction bar. |
| Shimmer | An exciter with its own dynamic control. Positive: new harmonics are generated from the 3 to 7 kHz region and added above the character's corner (Silk 8 kHz, Air 10 kHz, Glass 6 kHz), so a dull voice gets air it did not have. Negative: the air band is pressed down when it stands out. In both directions the band is kept in check dynamically. |
| Reverb | Tight (about 1 s), Large (about 3 s), Chamber (about 2 s, brighter, more motion). Fader is the wet level; -10 is silent. |
| Delay | Slap (1/16), Eighth (1/8), Quarter (1/4), synced to the song tempo, with a dulled feedback path. Fader is the wet level. |
| Width | A micro-pitch doubler: Stereo (±6 cents), Wide (±12 cents), Spreader (±9 cents with offset delays and extra width). Fader is the wet level. |
| Output | ±10 dB with a peak meter. FX duck under it: 0 to 12 dB. The reverb, delay and width sum is pulled down by this much while the vocal is present and released in the gaps, so effects bloom between phrases. |
Header
BYPASS is level-matched. The plugin reads the loudness difference between input and output at the moment you press it and plays the raw input at that level, and shows the match in dB next to the button. Press it with the chain settled, not during a gap.
Working with it
- Start with everything on its first character and all faders at 0. That is a working vocal chain already; move faders from there.
- The gain-reduction bars tell you whether De-ess, Compress and Shimmer are acting at all before you judge their sound.
- For a tuning pass, switch one module on at a time with the others Off and use BYPASS for the reference.
- The compressor sits before everything else that matters, so set it first; it decides the level the other modules see.
Under the hood
The compressor's clip stage and the exciter run at twice the sample rate behind a 63-tap half-band filter, so their harmonics do not alias; image rejection is about 110 dB and the response is flat to 19 kHz at 44.1 kHz. The price is a fixed latency of 31 samples (0.6 ms at 48 kHz), reported to Logic, which compensates it. The bypass path is delayed by the same amount so A/B stays aligned. Everything else in the chain is zero-latency.
The de-esser and the shimmer split the signal with Linkwitz-Riley 4th-order crossovers, which sum back exactly. The wet faders fade to silence on a cosine curve below 0, so the first steps are gentle.