Augmented Voices

How it works

Recorded voices + synthesis

Augmented Voices starts from recordings of solo singers and a chamber choir (dozens of hours of sessions), with over 60 articulations including sustained vowels, hums, attacks and experimental sounds, plus processed and extra voices. Alongside the samples work synthesis engines: virtual analog, granular (which breaks the voice into grains), harmonic (additive, ideal for 'artificial' choirs) and wavetable.

Two layers, four parts

The sound is divided into Layer A and Layer B, each with two parts (Sampler or Synth), one envelope per part, a filter (including a Formant filter that shifts the vowels) and two effects. The Morph knob gradually moves from one layer to the other, for example from the real choir to a synth pad, acting on several parameters together.

Harmonies with one key or voice by voice

The engine is polyphonic (up to 16 voices): you can play the harmonies by hand, note by note, or use Chord Select to make each key play a preset or custom chord. The first approach gives real voice leading, the second is a quick sketch in which all the voices move in parallel.

Character and tone

A hybrid vocal sound: the acoustic layer brings breath, vowels and human vibrato (recordings of solo voices and a chamber choir, male and female, with close and far microphones); the synth layer can add a pad, a sub, digital brightness or turn the voice into grains and textures. The real voice mostly occupies 100 Hz–8 kHz: male fundamentals between 80 and 400 Hz, female ones between 160 and 1000 Hz, the formants that define the vowel between 300 Hz and 3 kHz, 'presence' and sibilance between 3 and 8 kHz. Strengths: speed in sketching believable choral harmonies, Morph between choir and synth, Formant filter to change the vowel, arpeggiator with chords. Limits: no lyrics (only vowels and articulations), maximum polyphony of 16, very exposed solo choirs can reveal the sampled nature, and the one-key chord (Chord Select) moves all the voices in parallel and does not replace real voice leading. Compared with the Mellotron V choirs it is modern, clean and modulatable; compared with a choral library with lyrics (word builder) it is more immediate but less 'sung'.

Architecture

Signal path

• Layer A and Layer B with 2 slots each: Sampler (Chamber Choir, Solo Voice, Experimental & Processed; male and female voices; close and far microphones) or Synth (Analogue, Granular, Harmonic, Wavetable; the newer versions also include a Simpler engine).
• Each part → ADSR envelope with curves, Time Mult, Vel>Time → layer filter (SEM, Multimode, Surgeon, Comb, Phaser, Formant) → 2 insert FX → Volume/Pan → Morph → Main FX (digital, pitch shift or tape Delay → digital or convolution Reverb with 29 spaces) → output.
• Modulation: 2 LFOs, 2 22-point Functions, 2 Randoms (Turing / SnH), Vibrato and Tremolo assignable to the layers, Keys sources (velocity, aftertouch, mod wheel, keyboard tracking), MPE.
• Arpeggiator with 16 steps and Chord per step, Chord Select (11 chords or custom chord, active even with the arpeggiator off), Hold, swing and octaves.
• Macros: Morph (8 destinations), Color, Time, Motion (6 each), FX A, FX B, Delay, Reverb (4 each).
• Polyphony from mono to 16 voices, mono legato and glide; over 500 factory presets; MTS-ESP microtuning.

The controls, one by one

Main Panel — sound macros

Morph (large knob)

Blends Layer A and Layer B; it can control up to 8 parameters across the two layers at once (volumes, filters, wavetable positions, effects). Typically the acoustic sound is on the left and the synthetic one on the right.

Color

Color and harmonic content: acts on engine and filter parameters of each layer (up to 6 destinations). Turning it darkens or brightens the timbre.

Time

Preset timings: acts on the layer envelopes (up to 6 destinations). Toward the left, short attacks and releases; toward the right, long and soft ones.

Motion

Movement: controls up to 6 parameters of the Modulation and Arpeggiator pages (LFO speed and depth, functions, arpeggio).

Main Panel — effects macros

FX A

Layer A insert parameters (up to 4).

FX B

Layer B insert parameters (up to 4).

Delay

Master delay parameters (first effect in the master chain), including dry/wet.

Reverb

Master reverb parameters (after the delay), including dry/wet.

Advanced — structure and Layers

Navigation

Layers, Modulation, Arpeggiator, Effects, Macros pages; at the bottom, the modulation strip shows sources, MPE and macros in real time and lets you assign them by dragging onto controls (the destinations light up).

Layer A / Layer B — Volume and Pan

Level and stereo position of each layer.

Slots 1 and 2 per layer

Each layer contains two parts, each with a Sampler or Synth engine: up to 4 parts in total (e.g. 2 samples + 2 synths).

Advanced — Sampler engine (voices)

Sample selection

Library of real voices (chamber choir and solo voice, male and female, with dozens of articulations), processed voices and experimental sounds.

Volume

Sample level.

Coarse

Tuning in semitones: useful for octaving a sample (e.g. +12) or creating fixed intervals between parts.

Fine

Fine tuning in cents: 3–8 cents between two similar parts create a wider ensemble.

Random Start

Randomizes the sample start point: each note starts at a different point, avoiding the 'machine gun' effect on repeated notes.

Width

Stereo width of the sample: narrower for a solo voice in the center, wider for a choir.

Advanced — Synth engine: Analogue

OSC 1, 2, 3

Waveform of each oscillator: sine, triangle, saw, inverse saw, square.

OSC 2 and 3 Level

Levels of oscillators 2 and 3 relative to 1.

Noise

Noise generator: adds 'air' or breath to the acoustic sound.

FM

Frequency modulation from a source oscillator to VCO1 and/or VCO2: metallic harmonics.

Detune

Detuning between unison voices: widens and thickens.

Coarse / Fine

Tuning ±2 octaves in semitones and in cents.

Volume

Engine level.

Advanced — Synth engine: Granular

6-slot window

Six sample slots as starting material.

Start

Point in the sample where the grains start.

Density

Grain generation rate, in Hz or synced (straight, dotted, triplet).

Size

Grain length: time (1 ms–1 s), synced values or ratio relative to Density.

Env Shape

Grain window: from soft to 'clicky'.

Coarse / Fine / Volume

Tuning and level.

Advanced — Synth engine: Harmonic (additive)

FM

Frequency modulation on the fundamental, which propagates to the whole harmonic series.

Section

Window of the harmonic series on which the spectral filters act.

Morph

Crossfade between two factory spectral filters (different from the Morph macro).

Parity

Proportion of odd to even harmonics: all odd = hollow, clarinet/square-like timbre, all even = a more 'octaved' timbre.

Partials

Number of partials: few = soft, many = bright.

Coarse / Fine / Volume

Tuning and level.

Advanced — Synth engine: Wavetable

Position

Scans through the wavetable's waves (2D or 3D view); an excellent destination for LFO and Morph.

FM

Frequency modulation on the wavetable.

Phase Distortion

Warps the waveform over time: a sharper sound.

Wavefolding

Folds back the wave peaks, adding harmonics.

Detune

Detuning between unison voices.

Coarse / Fine / Volume

Tuning and level.

Advanced — Amplitude Envelope (for each of the 4 parts)

Attack / Decay / Sustain / Release

Amplitude envelope of the part (also by dragging the points on the graph); the link joins Decay and Release.

Att Curve / Dec Curve

Curvature of attack and decay: linear, exponential or logarithmic.

Time Mult

Multiplies or reduces the duration of the whole envelope.

Vel>Time

With high velocity the attack gets shorter, with low velocity it gets longer: useful for sounds that come in slowly when played softly.

Advanced — Filter (one per layer)

SEM (12 dB/oct)

Filter inspired by the Oberheim SEM, in lowpass, notch, bandpass or highpass: Cutoff 20 Hz–20 kHz, Resonance, Env Amt.

Multimode (24 dB/oct)

Arturia analog filter in the same four modes: Cutoff, Resonance, Env Amt.

Surgeon (64 dB/oct)

Extremely steep filter: Cutoff, Spread (width of the zone in notch or bandpass), Env Amt.

Comb

Comb filter: Frequency, Gain, Env Amt, Feedback/Feedforward.

Phaser

Phase-shifted peaks and notches: Cutoff, Feedback, Env Amt, Poles 4/6/8.

Formant

Vocal filter: Freq Shift (shifts the formant region), Q (emphasis), Morph (moves from one vowel to another).

Off

The filter can be turned off.

Modulation — LFO 1 and 2

Unipolar

Positive-only output (offset from the base value).

Polyphonic

One cycle per voice (not available in free mode).

Play Modes: Env / LFO / Key Trig

A single cycle on key press, continuous cycle, or continuous but restarted on every key.

Shape / Rate / Rate Sync

Waveform and speed, in Hz or synced (straight, dotted, triplet).

Fade / Phase / Amp

Fade-in, starting phase and output amplitude.

4 destination slots

Under each LFO, four destinations with amounts.

Modulation — Vibrato and Tremolo

Assignment

Each can be assigned to Layer A, Layer B or neither.

Fade / Rate / Depth / Poly

Fade-in on each voice, speed in Hz, depth and polyphonic mode (each voice with its own cycle).

Modulation — FUN (Function 1 and 2)

Graphic display

Drawable envelope/LFO with up to 22 points: click adds, right-click removes, arrows to curve; Point, Level and Time for fine adjustment; shape presets can be saved.

Drawing tools

Free, Line, Ramp, Sawtooth.

Play Modes

Env (once, with a sustain point settable by double-click), LFO (cycle), Key Trig (restarted cycle).

Scale / Rate / Rate Sync / Bipolar / Polyphonic

Amplitude, free or synced speed, bipolar output, cycle per voice.

Modulation — RND (Random 1 and 2)

Turing mode

'Controlled' random sequence: Length (sequence length), Flip (probability of inverting the peaks).

SnH mode

Sample and Hold: Smooth rounds off the steps.

Rate / Rate Sync / Amp / Polyphonic / Play Modes

Speed, sync, amplitude, per-voice behavior and Env, LFO, Key Trig modes.

Modulation — KEYS and Keyboard Settings

Velocity / Aftertouch / Modulation Wheel / Keyboard

Keyboard sources with customizable curves (up to 4 points) and four destinations each; for example velocity on brightness, aftertouch on vibrato, keyboard tracking on Morph.

Glide (Portamento / Legato)

Glissando time between notes; Legato only on legato notes.

Pitch Bend

Range in semitones.

Velocity > Amp and Curve

Maps velocity to the volume of both layers, optionally with the curve from the Keys page.

Polyphony

Voice limit from mono to 16, plus mono legato.

Voice Steal (Reassign / Rotate)

Reassigns the oldest voice only when needed, or always rotates to the next voice.

Arpeggiator — Step Controls

Step (1–16)

Turns each step on or mutes it without changing the pattern length.

Chord

On the selected steps, plays the held chord or the one set in Chord Select.

Velocity Bars / Gate Bars

Velocity and length of each step.

Randomize / Polyrhythm / Reset

Random variation of velocity or gate, different lengths for the two rows, reset.

Arpeggiator — Playback Controls

Mode

Up, Down, Up/Down Exc, Up/Down In, Random, As played.

Division and Rate Type

Speed in free BPM or synced (straight, dotted, triplet).

Swing

Delays the second sixteenth note of each eighth note.

Octave

Octave range of the arpeggio.

Chord Select and Chord

Chooses one of 11 preset chords or a custom chord, played from a single key; it also works with the arpeggiator off.

Hold

Holds the notes after release, even without the arpeggiator.

Effects — Layer FX (2 per layer)

Multi Filter

Resonant 12, 24 or 36 dB/oct filter in lowpass, bandpass or highpass.

Parametric EQ

5 bands: low and high shelving and three parametric bands with Q.

Compressor / Distortion / Bitcrusher

Dynamics, saturation, sample-rate reduction (up to 80×) and bit reduction (from 16 to 1.5).

Stereo Chorus / Juno 6 Chorus / Stereo Pan / Phaser / BL20 Flanger

Modulations: chorus with up to 3 voices, chorus modeled on the Juno-6, autopan, phaser, flanger.

Reverb / Digital Delay / Pitch Shift Delay / Tape Delay

Space and delays for each individual layer (the Pitch Shift Delay transposes the repeats).

Effects — Main FX

Delay (Digital, Pitch Shift, Tape)

Time and Time Sync, HP Freq and LP Freq on the delayed signal only, Feedback, Stereo, PingPong, Volume; the Pitch Shift adds Pitch-Shift and Offset between channels.

Reverb (Digital or Convolution)

Common: Input HP, Input LP, Decay, Size. Digital: Damping and Pre-Delay. Convolution: Input Gain and Room Type with 29 spaces (churches, halls, imaginary spaces).

Macros page

Sound (Color, Morph, Time, Motion)

Up to 6 destinations per macro (8 for Morph), each with On/Off, Amount and Curve. The sound macros can in turn be modulated by velocity, aftertouch, mod wheel and keyboard tracking.

Effects (FX A, FX B, Delay, Reverb)

Up to 4 destinations per macro with On/Off, Amount and Curve.

Workflow: sketching choral harmonies

One track per section (S, A, T, B)

The most realistic method: four instances (or four MIDI tracks on different instances) with female samples for sopranos and altos and male for tenors and basses, Polyphony at 1–2 voices per instance. Each section plays its own line, so the voice leading is real and you can export the parts to a score.

One instance, chords played by hand

Faster: a mixed chamber choir, Polyphony 8–16, and four-voice chords played with both hands (bass and tenor in the left, alto and soprano in the right). Check the ranges: soprano about C4–G5, alto F3–D5, tenor C3–G4, bass E2–C4.

Chord Select for the sketch

Set a custom chord (e.g. root, fifth, tenth) and play the bass melody with one finger: all the voices follow in parallel. It is perfect for trying out progressions and moods in seconds, but then it needs to be 'rewritten' with separate voices to avoid parallel fifths and octaves.

Hold for drones

With Hold on, sustain a pedal note (e.g. the bass on the tonic) while you try out the other voices above it: useful for immediately hearing suspensions and resolutions.

Time and Attack for believability

A real choir does not attack instantly: 100–300 ms of attack and 400–900 ms of release make the chords sound singable; for rhythmic stabs shorten everything with the Time macro.

Mod wheel as dynamics

Assign the mod wheel or aftertouch to volume and Color (or to the cutoff): choirs live on crescendos and diminuendos over long notes, more than on velocity.

Formant filter for the vowel

On a synth or processed layer, the Formant filter's Morph shifts the perceived vowel (from a dark 'oo' to an open 'ah'); automate it to make the chords 'speak'.

Sound design tips

• Ranges and tessituras: write within the registers of real voices (soprano C4–G5, alto F3–D5, tenor C3–G4, bass E2–C4); beyond these limits the samples sound strained or unnatural and the listener notices immediately.
• Spacing: keep at most an octave between soprano and alto and between alto and tenor; between tenor and bass there can be more space. Close chords in the low register (thirds below C3) turn to mud.
• Voice leading: hold common tones, move the other voices by step, avoid parallel fifths and octaves between the same two voices, preferably double the root in root-position triads and do not double the leading tone. Chord Select violates these rules by nature: use it for the sketch, not for the final version.
• One timbre per section: female samples on S and A, male on T and B; a slight difference in Fine (2–5 cents) and in panning (S and T slightly left, A and B slightly right, or the classic layout) gives depth without effects.
• Frequencies in the mix: high-pass at 80–100 Hz on female voices and 50–70 Hz on male ones; cut 2–3 dB at 300–500 Hz if the choir is 'nasal' or muddy; gentle presence at 2–4 kHz for vowel intelligibility; de-esser or a cut at 6–8 kHz if the breathy samples are sibilant.
• Hybridize with purpose: the synth layer should support (sub below 150 Hz, pad behind the choir, granular for the tail), not compete in the 300 Hz–3 kHz zone where the formants live.
• Vibrato: recorded voices already have a natural vibrato; use the engine's Vibrato sparingly (or on the synth layer only) to avoid an unnatural 'double vibrato'.
• Attacks and breath: varied Random Start keeps repeated notes from sounding identical; a slightly slower attack on low voices than on high ones imitates a real choir.
• 16-voice polyphony: four sections with long releases use up voices quickly; if notes get cut off, reduce the Release or split the sections across different instances.
• Reverb: for a believable choir use Pre-Delay 20–40 ms and decay 2–3.5 s (church or hall); for ambient, large spaces only on the synth or granular layer.
• Common mistakes: full chords in a single register, random doublings of the third, too much stereo effect that 'scatters' the choir, zero attacks on slow chords.

Presets to recreate

In genres and records

• Soundtracks, trailers, video games: epic, mysterious or liturgical choirs, often hybridized with synthesis.
• Ambient, new age, neoclassical: ethereal vocal pads and drones.
• Pop, dream pop, indie: background 'ooh' choirs and wordless vocal hooks.
• Electronic (melodic techno, trance, progressive, downtempo, future garage): granular vocalises and rhythmic vocal chords.
• Cinematic hip hop, trap and drill: dark choirs as a bed.
• R&B, gospel, contemporary worship: supporting choirs for sketching harmonies before the session with the singers.
• Sacred and choral music: study and sketching of four-part harmonizations.
• Synthwave and retro: 1980s-style synthetic choirs.