The part you were about to throw away

A reader sent me a short mail last weekend. They were working on a string part with a delay on it, the delay was building up resonances, and they put SMOOTH on to calm them down. Then they switched Delta on, liked what they heard, and bounced that version instead.

Delta was never meant to be listened to like that. Here’s why it works anyway.

What Delta plays

Delta plays back only what SMOOTH takes out: the dry signal minus the processed one. In the docs I say it should sound metallic, not musical, because when you use it as a check, musical means you’re cutting part of the song.

On a delay that turns around, but maybe not for the reason you’d guess.

A feedback delay does comb the sound. The peaks sit one divided by the delay time apart, so at 375 ms (a dotted eighth at 120 bpm) they’re 2.7 Hz apart. Far too close to hear as a tone. The comb only turns into a pitch at very short delays: at 5 ms the peaks are 200 Hz apart, and that’s the flanger sound.

So on a normal musical delay, the ringing comes from the strings themselves. The repeats stack the part on top of itself, and a chord held for two bars with the feedback up plays three or four times at once. The notes and their overtones pile up, the ones that stick out furthest above their neighbours get caught, and Delta hands you those on their own. It’s in key, it’s in time, and it follows whatever the strings play.

Intense makes it bigger. It multiplies Depth by three and lets SMOOTH cut up to 36 dB instead of 24, so more of the pile lands in the Delta signal. The reader said it worked with Intense on and off, and that matches the code: off gives you less, on gives you more.

There’s a Danish saying, man skal ikke skue hunden på hårene (don’t judge a dog by its fur). I’d filed Delta as a meter. Turns out it’s also a sound.

The bit worth keeping

Whatever a cleanup tool takes out of your track came from your track. That’s the takeaway, and it goes past SMOOTH: the part you were about to throw away is material you already made, and it fits the song better than anything you’ll pull from a pack. Print it before you decide.

In Live it takes a minute:

1. SMOOTH last on the string track, after the delay, Delta on.
2. Make a new audio track and set its input to the string track, Post FX.
3. Arm it, record a pass, then switch Delta off on the original.

Now the ringing is a clip. Chop it, pitch it, reverse it, or lay it quietly under the original part. Depth and Sensitivity decide how much ends up in that clip, so they act more like a tone control than a cleanup knob.

Four things to try

Try Intense both ways. Same settings, one pass with Intense off and one with it on. Keep whichever sits better, or use both on different sections.

Set levels with the printed clip. In Delta mode SMOOTH skips its level compensation on purpose, so the delta isn’t loudness-matched to the normal output. Balance the recorded audio instead of the plugin flipping back and forth.

Try it on other things that ring. A pad through a long reverb, a piano with the sustain pedal down, a vocal’s esses through a de-esser with a listen switch. Anything that builds up has something to print.

The normal use still holds. On a vocal or the mix bus, Delta is still the check. If what comes out of it sounds like the song, back Depth off.

From the studio

Still in the lab most days, mostly on the big update for the four. This issue came from someone using SMOOTH in a way I never planned for.

So that’s the ask this week: if you’ve found a use for any KERN plugin that isn’t in the docs, reply and tell me. This whole issue came from one reply, and I read all of them.

-Jonas

kernaudio.io