You load an automatic double tracking plugin on a vocal, pick a preset, and the track sounds wider. Then you push the modulation depth and the vocal starts swimming. Shorten the delay time and the effect collapses into chorus. Automatic double tracking occupies a narrow parameter window, and understanding that window determines whether you get a convincing synthetic double or a generic widening artefact that could have come from any modulation effect.
ADT produces a convincing "two performances" illusion only when the producer controls the specific parameter space that separates it from chorus, flanger, and slapback delay. Parameter literacy, not preset selection, is the difference between effective and ineffective automatic double tracking.
What automatic double tracking actually does
Automatic double tracking (ADT) is a studio technique that makes a single recorded performance sound like it was performed twice. It works by duplicating the signal and applying continuously varying time and pitch offsets to the copy, creating what the listener perceives as a second, independent performance rather than a processed version of the first.
The practical problem ADT solves is straightforward: you have one take of a vocal or instrument, and you want it to sound wider, thicker, and more present, as if two performers played the same part. Recording a genuine second take is the gold standard, but it requires a performer who can deliver a consistent yet naturally varied repetition. ADT achieves a similar perceptual result from a single recording.
ADT is distinct from a simple delay, which produces a static, audible echo. It is distinct from chorus, which uses shorter delay times and faster modulation to create a shimmering, ensemble-like effect rather than a doubling illusion. And it is distinct from a flanger, which operates at very short delay times with feedback to produce comb filtering. ADT sits in a specific parameter region between these effects. If you want a simpler overview of what ADT is and when to use it, see what is automatic double tracking.
How ADT works: the signal chain
The core ADT signal path is: the dry original signal, plus a delayed copy whose delay time is continuously modulated. That modulated delay is the mechanism. Everything else in ADT follows from how three parameters are set.
Delay time: 20 to 50 milliseconds. This is the Haas effect zone. Delays below roughly 20ms tend to produce comb filtering or chorusing rather than a perception of two distinct sources. Delays above roughly 50ms begin to sound like a distinct echo or slapback rather than a doubled performance. Within the 20-50ms range, the human ear perceives the delayed copy as a separate source arriving from a different position, not as a repetition of the original.
Modulation: slow, randomised variation of the delay time. A static delay at 35ms sounds like a fixed offset, not a second performer. What makes ADT convincing is that the delay time is never constant. It drifts slowly and randomly, typically by a few milliseconds in either direction, simulating the natural timing inconsistency between two human performances. No two phrases in a real double are ever perfectly synchronised; the modulation mimics that drift.
Pitch variation: 5 to 15 cents of detuning. When a singer performs a part twice, the pitch of the second take is never identical to the first. Subtle pitch differences, typically in the range of 5 to 15 cents, contribute to the perception of two independent voices. ADT applies this variation to the delayed copy, either through direct pitch shifting or as a natural consequence of modulating the delay time (changing the speed of a delay line shifts pitch, the same principle the original tape-based ADT exploited).
The original Abbey Road implementation made this principle physical. Ken Townsend's 1966 system used a second tape machine whose capstan speed was controlled by an external oscillator. As the oscillator varied the capstan speed, the second machine's playback drifted in time and pitch relative to the first, producing a continuously varying delay with inherent pitch modulation. Modern ADT plugins achieve the same result digitally, but the underlying mechanism, a modulated delay with pitch variance, is unchanged.
The history of ADT: Abbey Road, 1966
ADT was invented by Ken Townsend, a technical engineer at Abbey Road Studios, in 1966. The problem he solved was specific: John Lennon disliked the tedious process of manually double tracking his vocal parts and wanted a way to achieve the doubled sound without singing everything twice.
Townsend's solution used a second tape machine connected to the studio's mixing desk. The original vocal signal was fed to both machines simultaneously. The second machine's capstan motor was driven by a separate frequency oscillator rather than the studio's fixed mains frequency, which meant its playback speed could be varied in real time. As the oscillator modulated the capstan speed, the signal from the second machine arrived at the mix position with a small, continuously changing delay relative to the original. The speed variation also introduced natural pitch drift, exactly the combination of time and pitch offset that creates a convincing doubling effect.
The technique was first used extensively during The Beatles' Revolver sessions in 1966, and then throughout Sgt. Pepper's Lonely Hearts Club Band the following year. Lennon reportedly referred to the process as "flanging," a term that stuck within the studio and eventually became the name for the flanger effect, a related but distinct process that operates at shorter delay times with feedback.
ADT spread rapidly through the recording industry after 1966. Other studios built their own tape-based ADT systems, and the technique became a standard vocal processing tool well before digital implementations made it accessible to every producer with a DAW. The history matters because it explains why ADT sounds the way it does: the original design was shaped by the physics of tape machines, and the parameters that define effective ADT today (modulated delay, pitch drift, randomised variation) are direct inheritances from that physical process.
ADT vs. manual double tracking vs. chorus
These three techniques occupy different points on a single parameter continuum, and understanding where each sits clarifies when to reach for ADT specifically.
Manual double tracking is the practice of recording the same part twice and playing both takes simultaneously. It produces the most convincing doubling effect because the variation between two real human performances is complex and multidimensional: timing, pitch, dynamics, articulation, and timbre all differ independently between takes. No synthetic process replicates that full complexity. The trade-off is that manual double tracking requires a performer who can deliver a consistent second take, and it doubles the recording time.
Automatic double tracking creates a synthetic "second take" from a single recording. It is faster than manual doubling, fully controllable after recording, and works even when the original performance cannot be repeated. The limitation is that ADT's variation is less complex than real human variation: it operates on two dimensions (time and pitch) rather than the five or more dimensions that differ between two real performances. The result is convincing but identifiably different from a real double, particularly on exposed, dry sources.
Chorus uses the same underlying mechanism as ADT (a delayed copy with modulated delay time) but operates in a different parameter region. Chorus typically uses shorter delay times (below 20ms), faster and more regular modulation, and is often applied with multiple voices. The perceptual result is a shimmering, ensemble-like thickening rather than a "two people playing the same thing" illusion.
The continuum runs from manual double tracking (longest effective delay equivalent, maximum natural variation) through ADT (20-50ms, slow random modulation) through chorus (shorter delay, faster modulation) to flanger (shortest delay, feedback, comb filtering). Moving along this continuum, delay time decreases, modulation rate increases, and the perceptual result shifts from "two separate performances" to "one performance with a modulation effect." ADT sits in the specific region where the illusion of a second performer holds. Step outside that region in either direction and you get a different effect entirely.
For a fuller treatment of double tracking as a broader production technique, including manual doubling methods and when to use them, see double tracking.
Using ADT in practice: vocals, guitars, and beyond
Vocals are the most common ADT application. ADT on a lead vocal adds width and presence to a centred performance without requiring a second take. Typical vocal ADT settings use delay in the 25-40ms range with gentle modulation and 5-10 cents of pitch variation. The processed signal is often blended at a lower level than the dry vocal (around 50-70% wet) to maintain the original performance's clarity while adding the doubled width.
Rhythm guitars respond well to ADT, particularly when panning the dry and processed signals to create stereo spread. ADT on a distorted guitar can thicken the sound without the tonal variation that comes from recording a second take through a slightly different amp tone. Be cautious with heavily distorted sources: the harmonics in distorted signals interact with the pitch variation, and excessive detuning can produce an unpleasant beating effect.
Synths and keys can benefit from ADT when the source needs width without the cyclic modulation of a built-in chorus or unison effect. However, if the synth patch already includes significant chorus or detuning, adding ADT on top tends to compound the modulation into a washy, unfocused sound. ADT works well on dry, centred synth patches that need spatial width.
Mono compatibility is a critical concern with ADT. When the dry and processed signals are summed to mono (which happens on mono playback devices, in mono-summed club systems, and in the centre image of stereo playback), the delayed signal combines with the original, creating comb filtering. This produces audible thinning or hollowness in the summed signal. Check every ADT application in mono before committing. If the mono sum sounds noticeably thinner or phasey, reduce the wet level or narrow the delay time range.
Three common mistakes account for most ineffective ADT:
- Too much modulation: the delayed copy drifts too far in time and pitch, and the vocal sounds seasick or wobbly rather than doubled
- Too long a delay: the delay exceeds the Haas zone and the listener hears a slapback echo rather than a second performance
- Too much pitch shift: the detuning exceeds the range of natural human pitch variation and the result sounds like a detuned unison rather than two voices
All three mistakes share a root cause: leaving ADT's parameter window. The technique works within specific ranges. Outside those ranges, it becomes a different effect.
ADT in the modern DAW
If you want to understand what a dedicated ADT plugin is actually doing, or if you want ADT without purchasing additional software, you can build a functional ADT chain from stock DAW tools.
The DIY approach: Insert a delay on the track (or a send), set the delay time to 20-50ms with no feedback, apply an LFO to modulate the delay time slowly (rate below 1 Hz, depth of a few milliseconds), and add subtle pitch shifting of 5-15 cents to the delayed signal. This produces the core ADT effect: a modulated delayed copy with pitch variance.
The limitation of the DIY approach is predictability. A stock LFO modulates in a cyclic, periodic pattern, which can sound mechanical on sustained notes. Real human timing variation is random, not periodic. Dedicated ADT plugins address this by using randomised modulation algorithms that avoid the cyclic repetition of a simple LFO.
Dedicated ADT plugins add several capabilities beyond what stock delay tools provide: randomisation algorithms that produce more natural, non-cyclic variation; tape emulation that models the saturation and frequency response of the original analogue ADT systems; and in some cases, formant-aware processing that treats different frequency regions of a vocal independently. For a guide to specific ADT plugins and how to choose between them, see automatic double tracking plugins.
Routing affects the character of the result. ADT as an insert processes the signal directly and outputs a wet/dry blend. ADT on a send allows parallel processing: the original signal passes untouched while the doubled copy is blended in on a separate fader. The send approach gives more control over the blend and makes it easier to EQ or process the doubled signal independently.
Automation is where ADT becomes a production tool rather than a static effect. Using more ADT on choruses and reducing it on verses creates dynamic width contrast. Automating the wet/dry blend, or muting the ADT send, gives the doubled vocal a sense of lift in high-energy sections without making the verses feel artificially wide.
The parameter space that defines automatic double tracking
Automatic double tracking works because it occupies a specific, narrow region on the delay-time/modulation continuum. Shorten the delay below 20ms and the effect becomes chorus. Lengthen it beyond 50ms and the listener hears slapback. Speed up the modulation and the sound shimmers rather than doubles. Remove the modulation and the result is a static offset, not a convincing second performance.
The three parameters that define effective ADT are delay time (20-50ms), modulation rate and depth (slow, random, small excursion), and pitch variance (5-15 cents). Presets in ADT plugins set these parameters to reasonable defaults, but they cannot account for the source material, the arrangement density, or the specific sonic character you are targeting. Knowing what each parameter does, and what happens when you move outside the effective range, is what separates intentional use of ADT from generic preset application.
Automatic double tracking (ADT) simulates the sound of a performer playing or singing a part twice by duplicating the signal and introducing small, continuously varying time (20-50ms) and pitch (5-15 cents) offsets. Invented at Abbey Road Studios in 1966 by Ken Townsend, ADT remains a standard production technique for adding width and presence to vocals and instruments. The technique works within a narrow parameter space; outside that space, it becomes chorus, flanger, or slapback delay rather than a convincing double.