The Air preset in the PS-VC1 covers 3 kHz to 20 kHz, and sibilance on most voices lives between 4 and 10 kHz. Those two ranges are not neighbors. One sits entirely inside the other.
That is the part nobody warns you about when they tell you to add air to a vocal. Air is usually shorthand for a gentle shelf somewhere above 10 kHz, well clear of the consonants, and it costs you nothing on a well-recorded take.
A band labeled air that reaches down to 3 kHz is a different instrument. Every decibel you add there lands on the S sounds as squarely as it lands on the breath and the sparkle you were after.
Now look at what the high-end section actually offers. Sharp covers 5 kHz to 6 kHz, which is the baritone and tenor de-essing target almost exactly, and Edge runs from 1.2 kHz to 10 kHz.
That Edge range matters more than it looks. Sibilance can drop as low as 1.5 kHz on certain microphone and voice combinations, and it does so often enough in home studios that the low edge of that band is not theoretical.
Silk spans 1.8 kHz to 18 kHz, which is most of the vocal spectrum above the fundamentals. None of the four shapes in Black Rooster's fixed high-end presets is narrow enough to add top end while stepping around the sibilant region, because none of them was designed to.
A de-esser only acts when sibilance is present. Set correctly, the gain reduction meter flashes three to six decibels on the loudest S sounds and barely moves on the quiet ones, which leaves the rest of the performance untouched.
The reduction target is modest for a reason. Strip the esses out completely and the vocal stops sounding like speech, because sibilance is part of how consonants read as consonants.
You also have to pick your tool type. A wideband de-esser pulls the whole signal down when it detects an S, while a split-band version attenuates only the frequencies above your chosen point, and the second behaves far better under a bright shelf.
Inside a fixed chain, that position does not exist. The high-end lift happens on module two, before compression, reverb, and echo, so an external de-esser can only sit in front of the whole plugin or behind all of it.
Behind is the worst option. By then, the boosted sibilance has been through a compressor that pushes consonants forward, and it has been printed into reverb tails and delay repeats that a de-esser downstream cannot separate from the dry signal.
In front is imperfect but workable. Control the esses before anything amplifies them, and the compressor sees a signal whose consonants are not spiking, which changes how the whole chain behaves rather than just the top end.
There is one move left inside the plugin. The faders run to plus and minus 15 decibels, so Sharp can cut as easily as it can boost, and pulling 5 kHz to 6 kHz down two or three decibels works as a crude static de-ess on a voice that is consistently harsh.
Combine the two, and you get somewhere useful. Handle the dynamic sibilance upstream, trim Sharp if the voice needs it, then reach for Silk or Air last, which is the only order that gives you brightness that does not ride on the S sounds.
Clip gain still beats all of it. Pulling the two or three worst syllables down by hand before any plugin touches them costs a minute and removes the problem at the source, which no detector can do as cleanly.
That is the part nobody warns you about when they tell you to add air to a vocal. Air is usually shorthand for a gentle shelf somewhere above 10 kHz, well clear of the consonants, and it costs you nothing on a well-recorded take.
A band labeled air that reaches down to 3 kHz is a different instrument. Every decibel you add there lands on the S sounds as squarely as it lands on the breath and the sparkle you were after.
Your air shelf and your de-esser want the same frequencies
Published de-essing targets are specific. Around 5 kHz for a baritone, 6 kHz for a tenor, 6.5 kHz for an alto, and roughly 7 kHz for a soprano, with a working range of a kilohertz or two on either side.Now look at what the high-end section actually offers. Sharp covers 5 kHz to 6 kHz, which is the baritone and tenor de-essing target almost exactly, and Edge runs from 1.2 kHz to 10 kHz.
That Edge range matters more than it looks. Sibilance can drop as low as 1.5 kHz on certain microphone and voice combinations, and it does so often enough in home studios that the low edge of that band is not theoretical.
Silk spans 1.8 kHz to 18 kHz, which is most of the vocal spectrum above the fundamentals. None of the four shapes in Black Rooster's fixed high-end presets is narrow enough to add top end while stepping around the sibilant region, because none of them was designed to.
Static bands hit every syllable; dynamic ones do not
A shelf or bell applies the same gain whether the singer is holding a vowel or spitting a consonant. That is the whole difference between EQ and de-essing, and it is why cutting sibilance with EQ alone dulls a vocal.A de-esser only acts when sibilance is present. Set correctly, the gain reduction meter flashes three to six decibels on the loudest S sounds and barely moves on the quiet ones, which leaves the rest of the performance untouched.
The reduction target is modest for a reason. Strip the esses out completely and the vocal stops sounding like speech, because sibilance is part of how consonants read as consonants.
You also have to pick your tool type. A wideband de-esser pulls the whole signal down when it detects an S, while a split-band version attenuates only the frequencies above your chosen point, and the second behaves far better under a bright shelf.
The right de-esser position is the one you cannot reach
Consensus on placement is unusually firm. De-ess before any additive high-frequency boost, so you are not asking one processor to remove harshness that another processor deliberately added a moment earlier.Inside a fixed chain, that position does not exist. The high-end lift happens on module two, before compression, reverb, and echo, so an external de-esser can only sit in front of the whole plugin or behind all of it.
Behind is the worst option. By then, the boosted sibilance has been through a compressor that pushes consonants forward, and it has been printed into reverb tails and delay repeats that a de-esser downstream cannot separate from the dry signal.
In front is imperfect but workable. Control the esses before anything amplifies them, and the compressor sees a signal whose consonants are not spiking, which changes how the whole chain behaves rather than just the top end.
There is one move left inside the plugin. The faders run to plus and minus 15 decibels, so Sharp can cut as easily as it can boost, and pulling 5 kHz to 6 kHz down two or three decibels works as a crude static de-ess on a voice that is consistently harsh.
Combine the two, and you get somewhere useful. Handle the dynamic sibilance upstream, trim Sharp if the voice needs it, then reach for Silk or Air last, which is the only order that gives you brightness that does not ride on the S sounds.
Clip gain still beats all of it. Pulling the two or three worst syllables down by hand before any plugin touches them costs a minute and removes the problem at the source, which no detector can do as cleanly.
