Tape delay in plain English
Before digital memory was cheap, the easiest way to store a sound for half a second was to record it. So a tape echo runs a short loop of magnetic tape past a row of heads: the record head writes the incoming sound onto the tape, and a playback head a few centimetres further along reads it back. The gap between those two moments is the delay. An erase head clears the tape before it comes round to the record head again.
Turn up the feedback and some of what the playback head reads is sent back to the record head, so the echo gets recorded again and comes round for another pass. Each pass makes another repeat, quieter than the last. Every delay works like that, and the delay lesson covers the controls they all share.
What sets tape apart is what happens on each trip round the loop. Tape can't hold the highest frequencies well, so a little treble goes missing every time. It distorts gently when the level is high, which adds saturation. The motor and the tape path never run at an exactly even speed either, so the pitch of the echoes wanders. A digital delay does none of this unless someone programs it in, and those imperfections are what people mean when they ask for tape delay.
Delay time, heads and tape speed
The time is the distance between the heads divided by the speed of the tape. Suppose the first playback head sits 4 cm after the record head and the tape moves at 19 cm a second, about 7.5 inches per second. The tape needs 4 ÷ 19 = 0.21 seconds to cover the gap, so the echo arrives 211 ms late. A head three times as far along, at 12 cm, gives three times the delay: 632 ms.
On an Echoplex you set the time by sliding the playback head along. The Roland Space Echo keeps its heads fixed and changes the motor speed instead, which is what its Repeat Rate knob does. Speed the motor up and every head's echo comes sooner; slow it down and they all come later, in the same proportion.
You can hear the speed change, too. While the motor slows down, the sound already on the tape is played back slower than it was recorded, so it drops in pitch, the way a record does when you put your finger on it. Speed up and the echoes rise. Producers sweep the time on purpose for swooping dives and rises, and most tape delay plug-ins copy the effect. Turning the Time knob in the lab while the loop plays does the same, and the lab keeps the tape between half and one and a half times its speed so the bends stay musical.
The sound of tape echo
Darker repeats
Every trip through the record and playback heads loses some treble. The first repeat is a little duller than the original, the second duller again, and by the fifth or sixth the echo has turned into a soft, muffled outline of the note. Your ears hear high frequencies fading as distance, so a trail of darkening echoes sounds like it's moving away from you instead of piling up on top of the mix.
In the lab the Tone knob is a low-pass filter on the playback, which means it runs once per pass. At 2.5 kHz the first repeat loses about 7 dB of its top end compared with its middle, and the fourth repeat loses about 10 dB. On the echo timeline the teal mark on each bar drops further below the top of the bar as the repeats get darker.
Saturation
Tape can only hold so much signal. Record it hotter and the tape stops following the waveform in a straight line and starts to round off the peaks, which adds harmonics, new frequencies at whole-number multiples of each note. The curve tape follows is close to symmetrical, so most of what it adds is odd harmonics, the 3rd above all, with a little of the even ones. The tape saturation lesson looks at that curve on its own.
On a tape echo the saturation stacks up, because every repeat is recorded onto the tape again and keeps the distortion it picked up on earlier passes. Driven moderately, it makes the echoes thicker and softer at the edges, so they blend with the track. At 14 dB of drive in the lab, a 220 Hz note comes back with a 3rd harmonic at −18 dB and a 2nd at −25 dB, and the harmonics screen shows those bars for the first and fourth repeats side by side.
Wow and flutter
No motor turns at a perfectly even speed, and a loop of tape that has been running for years stretches and wears. The small speed changes that result are called wow and flutter. Wow is the slow part, a drift you can follow, and flutter is the fast ripple on top of it. Because the speed changes, the pitch of the echoes changes with it, while the dry sound stays exactly in tune.
Pitch changes this small are measured in cents, where 100 cents is a semitone. At 40% in the lab the echoes drift up to about 12 cents sharp or flat, an eighth of a semitone, which you hear as a gentle wobble on the repeats while the dry sound holds the note. At 100% they reach nearly 30 cents. That's the seasick warble of a worn-out machine, and some lo-fi producers want exactly that.
Wow on a delay is close to vibrato, where an LFO moves the pitch up and down, and chorus uses the same idea with much shorter delay times. Tape's movement is less regular than either: it wanders instead of repeating one smooth cycle.
Hiss and wear
Tape also adds hiss, a soft high noise that builds up a little with every pass, and old loops drop out for a moment where the coating has worn thin. The lab leaves both out so the screens stay readable. Many tape delay plug-ins include them with a Noise or Age control, and a little noise under a long trail of echoes is part of the vintage sound.
Multiple playback heads
A tape loop can have more than one playback head, and each one reads the same tape at a different point. On a machine with three evenly spaced heads, head 2 plays the sound back at twice head 1's time and head 3 at three times. Turn on two or three heads and every note comes back as a short rhythmic pattern instead of a single echo, and with feedback those patterns overlap and build into dense, galloping trails.
In the lab, with head 1 at 125 ms and heads 1 and 3 switched on, the echoes arrive at 125 ms and 375 ms. At 120 BPM a beat lasts 500 ms, so those are a 1/16 note and a dotted 1/8, and the repeats lock to the groove even though you only set one time. Change the time and both heads move together, so a tape echo's patterns keep their shape at any speed.
The Roland RE-201 Space Echo has three playback heads and a mode selector that picks which of them you hear and whether its spring reverb comes in too. The Binson Echorec, which David Gilmour used with Pink Floyd, does something similar with several heads around a spinning magnetic drum in place of a tape loop.
Self-oscillation and runaway feedback
On a clean digital delay, feedback above 100% makes every repeat louder than the last, and the level climbs until it clips. A tape echo behaves differently. As the echoes get louder the tape saturates harder, and saturation lowers the gain of the loudest parts, so the build-up levels off. You get self-oscillation: a sustained roar made of the echoes feeding on themselves, darkened by the tape on every pass.
Dub engineers in 1970s Jamaica played tape echoes like an instrument this way. Lee "Scratch" Perry worked with a Roland Space Echo at his Black Ark studio, and the move you hear on dub records again and again is feedback turned past the point of runaway for a bar or two, then brought back down before it swamps the track. Sweep the time while it's running and the roar dives and climbs with it.
In the lab feedback goes up to 110%, and at 105% the echoes settle at about the level of the original, so turn your volume down before you try it. The delay feedback lesson covers feedback on a clean delay, where there's no tape to hold it back.
Reading the lab
The wide screen at the top is the echo timeline. The white bar at zero is the original, and every pink bar after it is one repeat, placed where it arrives and as tall as it is loud. The H1, H2 and H3 labels show where each head's first echo lands. The teal mark on each bar shows how much treble that repeat has kept: on top of the bar, the repeat is as bright as the original, and the further down the mark sits, the darker the repeat.
Below it, the harmonics screen shows what the tape adds to a 220 Hz note, with the first repeat in pink and the fourth in violet; odd-numbered harmonics are labelled brighter. The pitch drift screen draws four seconds of how far the echoes move from the original pitch, worked out from where the playback head is reading as the tape speed wobbles. The Right now panel describes the control you last moved, using numbers measured from the same tape.
The lab models a mono tape machine, which is how most tape echoes were built: both channels go onto one tape, and the echoes come back in the middle of the stereo picture while the dry sound keeps its width. Press A/B to switch the echoes off at matched loudness, so the only thing that changes is the tape. You can use any sound, by uploading a file or pressing Record for eight seconds, and it stays on your device.
Tape echo machines and plug-ins
The Echoplex, designed by Mike Battle in the early 1960s, used a tape cartridge and a playback head on a slider, so you set the time by moving the head. The EP-2 used valves and the later EP-3 was solid state. In Britain, Charlie Watkins built the Watkins Copicat. The Roland RE-201 Space Echo arrived in 1974 with fixed heads, a variable motor speed and a spring reverb built in, and you can hear it all over dub and reggae.
Original units are expensive and need new tape, cleaning and servicing, so most people use a plug-in or a pedal. Universal Audio's Galaxy Tape Echo and Roland's own RE-201 plug-in model the Space Echo. Soundtoys EchoBoy has tape styles among its delay types, Logic Pro's Tape Delay has flutter controls and filters on its repeats, and Ableton Live's Echo can add noise and wobble from its Character section. On a pedalboard, the Strymon El Capistan and the Boss RE-20 are both tape echo models.
You can get close with any clean delay that has a filter in its feedback loop. Put a low-pass at 2 to 4 kHz and a high-pass at 150 to 300 Hz on the repeats, add a saturation plug-in to the delay's return, and modulate the delay time very slightly. The tape delay vs digital delay comparison sets the two side by side.
How to set up a tape delay
This works in any DAW with any tape delay plug-in.
- Put it on a sendLoad the tape delay on a return track, set its mix to 100% wet and send the part to it. Every change you make then affects the echoes only, and the insert vs send lesson explains why that helps.
- Set the time to the songAt 120 BPM, 375 ms is a dotted 1/8 and 500 ms is a 1/4. On a vocal, pick a time that drops the echo into the gap after each phrase, so it stays out of the way of the next word.
- Choose how many repeats you wantStart with feedback around 30 to 40% for three or four audible repeats. Go higher for long trails behind a sparse part, lower for a single answer.
- Darken the repeatsBring the tone down until the echoes sit behind the dry sound instead of beside it, usually somewhere between 2.5 and 5 kHz.
- Add a little drive and wobbleUse enough drive to soften the edges of the echoes. On a vocal or chords, keep the wow and flutter to a few cents of drift so the echoes still sit in tune with the dry part.
- Set the send levelBring the send up until you can hear the echoes when you listen for them and miss them when you mute the return.
Starting settings for common sounds
These assume 120 BPM and a delay on a send. Start here and move each control until the echoes sit where you want them.
| Use | Time | Feedback | Tone | Drive and wobble |
|---|---|---|---|---|
| Lead vocal | 375 – 500 ms | 25 – 40% | 3 – 4 kHz | Light drive, slight wobble |
| Vocal slapback | 90 – 130 ms | 0 – 10% | 4 – 5 kHz | Light drive, no wobble |
| Electric guitar | 375 ms | 30 – 45% | 2.5 – 4 kHz | Moderate drive, slight wobble |
| Synth lead | 250 – 375 ms | 35 – 50% | 2 – 3 kHz | Moderate drive and wobble |
| Lo-fi keys or drums | 125 – 250 ms | 20 – 35% | 1.5 – 2.5 kHz | Heavy wobble, some drive |
| Dub throw | 375 – 750 ms | 60 – 100%, ridden by hand | 1.5 – 3 kHz | Heavy drive, some wobble |
A dub throw is a single word, snare hit or chord sent to the delay for a moment by turning the send up and straight back down. Automate the send, and the feedback too if you want the tail to swell, instead of leaving the delay on the whole part. For a quick single repeat right behind a voice, read the slapback delay lesson.
Common tape delay mistakes
Too much wow and flutter on everything
A few cents of drift make echoes sound like tape. Somewhere past 15 cents they start to sound out of tune against the dry part, which suits a lo-fi effect and fights a vocal or a chord part that's meant to sound clean. Turn it down until you only notice it when it's gone.
Leaving the repeats bright
Without the treble loss a tape echo is just a delay, and bright echoes crowd the same space as the dry sound. If the repeats are cluttering the mix, darken them before you turn them down. They'll often sit fine at the same level once they're duller.
Echoes that cover the next phrase
Long feedback on a busy vocal smears every line into the next. Use fewer repeats, pick a time that lands in the gaps, or have the echoes duck while the voice is singing with a ducked delay.
Runaway feedback with no safety net
Self-oscillation can jump in level in under a second. Keep a limiter on the return while you experiment, and automate the feedback back down instead of trying to catch it with the mouse.
Training your ears to hear tape
The quiz under the lab plays the loop through the tape with one trait switched on: darker repeats, wobbling pitch, saturated repeats, or none of them. The controls and screens are hidden and the levels matched, and you name what's happening to the echoes. Saturation is the hardest to pick out at first, so listen to the edges of each repeat for a fuzzy, thickened sound. Eight rounds take a few minutes, and your best score is saved in your browser.
Questions people ask
What is a Space Echo?
The Roland RE-201 Space Echo is a tape echo from 1974 with a loop of tape, three playback heads and a spring reverb in one box. A mode selector picks which heads you hear, with or without the reverb, and the Repeat Rate knob changes the motor speed and with it the delay time. Roland's own RE-201 plug-in and Universal Audio's Galaxy Tape Echo both model it.
Is tape delay the same as analogue delay?
No, though both are analogue. "Analogue delay" usually means a pedal such as the Boss DM-2, which passes the sound along a bucket-brigade chip (a chain of tiny capacitors) instead of recording it onto tape. They lose top end and distort a little on every repeat in much the same way, which is why they get grouped together, but only tape has wow and flutter from a motor and a moving loop.
Why does the pitch change when I turn the delay time on a tape delay?
Because the time comes from the tape speed. To make the echo later, the motor slows down, and everything already on the tape plays back slower and lower while it does. Most tape delay plug-ins copy this, and some let you turn it off; Ableton Live's Delay calls the tape-style behaviour Repitch.
What do wow and flutter mean?
Both are wobbles in the tape speed, which show up as wobbles in pitch. Wow is the slow part, a drift you can follow, and flutter is the fast part, a ripple that makes the echoes sound grainy and less solid. On a tape delay only the echoes have them, because the dry sound never touches the tape.
Should tape delay go on an insert or a send?
A send, almost always. Set the tape delay to 100% wet on a return track and send each part to it, so you can darken, saturate or automate the echoes without touching the dry sound, and several parts can share one delay. Use an insert when the echoes belong to one sound's character, like a synth lead with its own tape echo.
Can tape delay damage my speakers when it self-oscillates?
It can get very loud very quickly, so treat runaway feedback with care. Turn your monitors down before you try it and put a limiter after the delay. Plug-ins differ here: some stop just short of runaway and others hold it in with saturation, as the lab does.