INT-Reverse

⚠️ Nobody has listened to it yet

It works end to end, it has its full window, and it joined the pack in 0.7.0, so it installs together with the other five. The only thing left is the one thing no machine can do: somebody listening to it with a real voice. If it sounds off, say so.

How to use it, in three clicks

1. Put the plugin on the vocal track and press Listen. 2. Play the section once. The sound does not change at all: it is only recording. 3. Press Generate. In a few seconds it finds where each phrase starts and builds the swells. 4. Play it again: now the swell comes in before every phrase.

If you do not like it, Clear throws the recording away and you start over.

Moving any control other than Amount and Level rebuilds the swells by itself, and waits until you let go of the knob before it starts. While it rebuilds you keep hearing the previous ones. Those two are gains: they respond instantly and rebuild nothing.

What it does, in one sentence

That sound that swells out of nowhere and lands right when the line starts: the reverse swell: is made by hand in every DAW, and it is always the same five steps: cut the bit, reverse it, add reverb, reverse it back, and drag it into place. Ten minutes per line. INT-Reverse does it for every line of a take, in one go.

Why it stays in tune without you telling it anything

Because the swell is built out of the very line that is about to come in. It does not generate a new sound that then has to be tuned: it takes the beginning of that line, reverses it, runs it through the space you picked, and reverses it back.

It comes out in the key of the line by construction, every time. Nothing to tune, nothing to pick.

Why it adds no latency

Putting a sound before another one means, in real time, knowing that other sound before it plays. That is: the plugin would have to ask your DAW for two seconds of lookahead. That is unusable for recording and annoying for mixing, and it is exactly why the similar plugins released in 2026 make you press a button for every single line.

Here the take is already recorded, so the swell is computed once and then simply added where it belongs. Zero latency.

Where to put it

It is an effect: it goes in a mixer slot, on the vocal track. Not in the instrument rack.

There is nothing to route. The voice runs through the plugin, so it already knows what to duck against when it needs to get out of the way. That routing step is where everyone gets stuck in other plugins, and here it does not exist.

The two controls that matter

Level | how many decibels below the phrase the swell comes in

This is the control that decides how the effect sounds, and since 0.8.0 it is a number you choose.

And what makes that number mean something: the plugin measures your phrase and puts the swell exactly there, whichever space you picked and however much material it is using. It did not work that way before: the level fell out of an internal calculation that nobody had chosen, and that is why the effect sounded wrong no matter what you moved.

Amount | the trim

Amount: how much of that swell gets added. At the factory value (100 %) you hear exactly what *Level* says.

It is mostly there to be automated: keep the effect on some phrases and off on others without touching any other control.

The drawing: seeing what it did

Your take is at the top. Three things are drawn on top of each other:

• Your voice, in grey, exactly as you recorded it. That is the reference.
• The swells, in colour. Not an approximation: it is the waveform of what is going to play. Turn *Amount* up and they get bigger, because both waveforms are drawn at the same scale precisely so that comparison means something.
• A vertical line where each phrase starts, which is the exact point where its swell ends.

At a glance you can tell whether the plugin found the phrases you hear. If there is a line in the middle of a silence, or one missing where you did sing, that is the problem, not the sound.

Where the tail comes from

Under the grid there is a dropdown that says Tail from. It has three options, and it decides which controls do anything:

Places is the usual one, and what you get out of the box. The tail comes from one of the six spaces, the same ones as INT-Spaces, ordered from smallest to biggest: booth, bathroom, wood room, chapel, cathedral, stadium. You pick by looking, not by reading a list. The space is the rough length of the swell: a booth gives you a short one and a cathedral a huge one. The factory setting is the chapel, which is the most useful tail for a voice without going all the way to a cathedral.

Diffuse is a cloud with no walls. It does not imitate any place: you say how long it lasts, how dark it is and how wide it feels. It is for when you know what you want and no space gives it to you.

Plate is the dense, metallic tail of a studio plate. It has almost no low end, so it does not cover the body of the voice that comes next. It is the one for reverse hits in rap and spoken word.

When you pick either of the last two, the grid of spaces goes dim and the three tail knobs turn on, under More options. There are never two live controls saying different things about the same thing.

A project saved with an earlier version opens on Places, with the same space it had. It sounds exactly as before: nothing changes unless you change it.

Taking the swells away on their own

Export saves just the swell layer as a WAV, without your voice. Next to it there is *"or drag from here"*: you can drag straight into your DAW from there.

It is the layer alone and not the mix, on purpose. You already have your dry voice on its track; with the swells separate you can send them to another channel, compress them or automate them without touching the voice.

Both turn on only once there are swells built.

The fine controls

They live behind More options. You do not need them for it to sound good.

Length: fine-tunes the length, between 15 % and 100 % of what the chosen space lasts.

Character: dark to bright. In the middle it is the space exactly as physics gives it.

Duck: how much the swell gets out of the way while the voice sings, so they do not clash.

Advance: how many milliseconds before the phrase the swell ends. At 0 it lands right on it.

Material: how much of the start of the phrase is used to build the swell. At 0 the plugin decides, which is what we recommend: it looks for the end of the first syllable. Too little material gives a thin swell; too much mixes different notes and smears it.

And these three only do something with Diffuse or Plate. On Places they are off, because there the grid decides the tail.

Tail: how long the tail lasts, from half a second to six. Half a second is the minimum on purpose: any shorter and the swell starts sounding like the word it came from.

Darkness: how fast the highs die off. To the right the tail sits further back and softer; to the left it decays evenly and you notice it more.

Size: how wide and enveloping it feels. To the left it sits in the middle, to the right it opens out to both sides.