What is reverse reverb?

Reverse reverb in plain English

Normal reverb starts the moment a sound does and dies away after it. Hit a snare in a big room and you hear the crack, then the room ringing on for a second or two, loudest at the start and fading to nothing. The reverb lesson covers how that tail is built.

Reverse reverb flips that tail around in time. The quiet end comes first and the loud start comes last, so instead of fading out after the snare, the reverb fades in before it, getting louder and louder until it stops dead as the crack arrives. It's like hearing the sound get sucked in out of nowhere, and it's been used on vocals and drums since the days of tape.

The swell is made from the sound it leads into. A reverse swell before the word “night” is the reverb of “night”, backwards, so it has the same pitch and tone as the word and blends straight into it. That's the difference between reverse reverb and a generic riser or a reversed cymbal sample: it belongs to the note that follows.

Reading the lab

The cards above the screens switch one plate reverb between three ways of playing. Normal plays it the usual way round, and Reverse gives you the swell you'd make by printing the reverb and flipping it. The third card, Reverse preset, plays the backwards tail that a reverb plug-in can produce live, and that turns out to land somewhere very different.

The big screen shows the whole loop. The pale shape is the dry sound and the coloured shape is the reverb, at the levels you hear them. The lab starts on an acoustic guitar loop with about three seconds of silence in the middle, because a swell needs space to rise in, and a vocal that never stops gives it none. Watch that gap: in Normal the colour trails into it from the left, and in Reverse it climbs out of it towards the second part. The drawing is to scale, so only the loud end of a swell shows up; you'll hear more of it than you can see. Hover over the loop to compare the two levels at any point.

Underneath, the decay wave is what the reverb does with a single click, the white line in the middle. Everything to its left happens before the click and everything to its right after it. The amber bracket marks the part of the reverb within 30 dB of its loudest point, and that's the length the readouts call the swell or the tail. The lab renders the click through the same processing you hear and times the result.

A/B switches the reverb off and sets the dry sound to the same loudness, so you can hear what the reverb adds without the level fooling you. Pick another sound from the list, upload your own, or record eight seconds with your microphone; nothing leaves your device.

How reverse reverb works

Every reverb has an impulse response, the sound it makes from a single click. For a plate with a 2.5 second decay, that's a burst of dense noise that falls by 60 dB over 2.5 seconds, which is the RT60. Whatever you send into the reverb, the output is that response laid down again for every instant of the input, scaled by how loud the input was at that instant.

Reverse the audio and run it through the reverb, and each moment of the backwards audio sets off a normal tail that runs forwards in time from it. Reverse the result and every one of those tails is now backwards, ending exactly at the moment that caused it. The audio itself is back the right way round, so the original performance plays as it was, with its reverb coming before it instead of after.

A swell that ends on a word has to start a second or two before the word, and it's built from that word, which a reverb running in real time hasn't heard yet. That's why reverse reverb is printed: you render the reverb to audio, where the whole performance already exists, and the future is just a bit further along the timeline.

The lab gets the same result without the printing. It feeds the plate, with its impulse response turned backwards, a copy of the loop that runs ahead of what you hear by the length of that response. The sum comes out sample for sample the same as printing and flipping. It's all rendered ahead of time, so the drawings and the sound you hear are the same samples.

Reverse reverb presets and why they sound different

Plenty of reverbs have a setting called reverse. Convolution reverbs often have a button that flips the impulse response, and some algorithmic reverbs have reverse programs. These play a tail that gets louder instead of quieter, which is a lovely sound, but they're still running live, so the tail can only start when the sound does.

Pick Reverse preset in the lab and watch the decay wave. The swell begins at the click and peaks to the right of it, as far after the click as the tail is long. With a 2.5 second plate the reverb is loudest about 3 seconds after the word that caused it, by which time the singer is usually on the next line. It can sound good on pads and long notes, where there's still something playing when the swell arrives. On a vocal or a drum hit it makes a whoosh that belongs to nothing.

The printed version is the one people usually mean by reverse reverb, and it's what the rest of this page is about.

How to make reverse reverb, step by step

This works in any DAW. The examples use one word at the start of a vocal phrase, which is the most common place for it.

1. Copy the part onto a new trackDuplicate the word, phrase or drum hit you want the swell to lead into, onto an empty audio track. Leave the original alone.
2. Reverse the copyIn Ableton Live that's the Rev button in the clip's Sample settings. In FL Studio, tick Reverse in the sampler channel's settings or use Edison. In Logic, the audio file editor's Functions menu has Reverse.
3. Add a reverb at 100% wetA long plate or hall works best. Set the mix fully wet so you only print the reverb, and give yourself plenty of decay to work with, two or three seconds to start.
4. Print the reverb with its tailBounce, resample or freeze and flatten the track, and make sure the render runs past the end of the clip by the full length of the tail. Cut it off at the clip and you lose the start of the swell.
5. Reverse the printed reverbFlip the new audio the right way round. It now rises in level towards its end.
6. Line up the swell with the soundSlide the reversed print earlier. If the print started exactly where the reversed copy did, it's in the right place when its end lines up with the end of the original word. Check it on the waveform: the swell climbs until the first transient of the word, and the climb should finish right on it.
7. Trim, filter and set the levelCut or fade anything after the peak that clutters the word, take the low end out with a high-pass filter around 200 Hz, and bring the fader up until the swell is clear without competing with the word itself.

You can do step 3 on a send instead, printing the return track, which lets you try several reverbs on the same reversed copy before you commit.

Setting the length of the swell

The decay control decides how far ahead of the sound the swell begins. A longer decay means a longer tail, and backwards that's a swell that starts earlier and builds more slowly. As a rule of thumb the loud part of the swell lasts a little under half the decay. A tail that falls 60 dB in 2.5 seconds is 30 dB down after about 1.25 seconds, and the lab measures 1.09 seconds for its plate, which loses its top end a little sooner than its mids.

To time a swell to the song, work out the length of a beat: 60,000 divided by the BPM gives it in milliseconds, so a beat at 120 BPM is 500 ms and a bar of four beats is 2 seconds. A swell of one beat wants a decay of a little over twice that; in the lab, 1.2 seconds gives a swell of about 530 ms. The audible start of the swell is soft and gradual, so it doesn't need to be exact, but the end does: it has to land on the sound.

Watch where the swell starts, too. In a vocal, the space before a phrase is often the end of the previous one, and a 2 second swell can start right over the last word of the line before. Shorten the decay until the swell fits in the gap, or fade in its start after the previous word ends.

Leaving a gap before the sound

With the peak right on the word, the swell runs straight into it and the two blend. Move the swell 30 to 80 ms earlier and a short silence opens up before the word, so the swell rushes in, stops, and the word lands on its own. It's the reverse of pre-delay, and it does a similar job: the front of the sound stays clean.

The gap is a matter of taste, and it depends on the word. A soft vowel at the start of a word sounds natural with the swell running into it. A hard consonant like a k or a t sounds sharper with a little space in front of it. In the lab, the Gap control moves the swell earlier, and on the decay wave the empty strip left of the click is that gap.

Damping and the tone of the swell

In a real room the high frequencies die away faster than the mids, because air and soft surfaces soak them up. Reverbs copy this with a damping control. In a normal tail it means the reverb gets darker as it fades.

Played backwards, that order flips too. With heavy damping the swell starts dull and murky and the top end only joins in just before the sound, which gives it more of a rush. Little damping keeps it bright all the way, glassy and ghostly, though harsh on a sibilant vocal. The Right now panel measures how long the top end above 4 kHz lasts against the whole swell: at 40% damping it's there for about 850 ms of a 1.1 second swell, and at 85% for only the last 500 ms.

The reverb's filters change the tone of the swell too. Backwards low end sounds like a rumble coming towards you, and it builds up under the start of the word. A high-pass filter on the printed swell, somewhere between 150 and 400 Hz on a vocal, keeps the effect in the mids and highs where it reads clearly.

Where to use reverse reverb

Reverse reverb is an effect, so it works best in small doses. The lab puts a swell in front of everything in the loop, which makes it easy to hear; in a song you'd usually print it for one word, one hit or one moment.

The first word of a vocal

This is where you'll hear it most. A swell into the first word of a verse or chorus makes the vocal seem to appear out of the track, and it works best when the line before has finished, so the swell has a clear space to rise in.

Snares and crashes

A backwards plate swelling into the first snare of a chorus, or into the crash on its first beat, marks the change of section. Use a shorter decay than on a vocal, under a second or so, so the swell starts within the beat before the hit.

Transitions and risers

Print a long swell, 3 to 6 seconds of decay, from the first chord or note of the next section and play it on its own at 100% wet. It's a riser built from the music that follows, so it's already in the right key.

Whole lines, for atmosphere

Reverse reverb on a whole phrase gives every word its own swell, and the result is the eerie, sucked-in sound used in horror soundtracks and dark ambient music. It blurs the words, so it suits a backing layer better than the lead.

Common reverse reverb mistakes

Putting the end of the clip on the start of the word

The reversed print is longer than the word by the length of the tail, because it holds the reverb of the whole word. Drag its end to the start of the word and the swell lands a whole word too early. If the print started where the reversed copy did, its end belongs at the end of the word, and either way the point where the swell stops climbing belongs on the first transient.

Cutting off the tail when you print

If the render stops at the end of the reversed clip, the tail gets chopped, and the tail is the start of your swell. Extend the render by at least the decay time.

A swell that starts over the previous line

A long swell needs room. If it begins while the last phrase is still going, the two smear together. Shorten the decay or fade in the swell after the previous word ends.

Leaving the low end in

The bottom of a backwards reverb piles up under the start of the word and muddies it. Filter it out of the swell; the word itself still has all its low end.

Putting it everywhere

One swell into a chorus is an event, but a swell before every word turns into a wash, so save it for the moments that need an entrance.

Where reverse reverb came from

Before DAWs, the trick was done with tape. You turned the reel over so the recording played backwards, recorded echo or reverb from it onto a spare track, then turned the reel back the right way. The effect track now ran backwards, ahead of the sound it came from. Jimmy Page has said he worked out this reverse echo while recording “Ten Little Indians” with the Yardbirds, and he used it at the end of Led Zeppelin's “You Shook Me”.

The idea hasn't changed since. A DAW makes each step a click instead of a reel flip, and a convolution reverb makes it easy to see what's going on, because its impulse response is a recording you can look at and turn around.

Train your ears

The quiz under the lab plays the loop with the reverb in a hidden mode and asks which way round it is: a normal tail, a swell leading in, or the live preset's late swell. The levels are matched and the screens covered, so you have to listen to where the reverb sits against each sound. Eight rounds take a couple of minutes, and your best score stays in your browser.

Questions people ask

Keep going