Vocoder V

How it works

Two signals, one output

A vocoder has two inputs. The modulator is the signal that speaks (usually the voice, but also drums or a loop); the carrier is the sound that will be made to speak (here, an internal synth). The output is the carrier shaped by the spectral envelope of the modulator.

Filter banks

The modulator passes through a bank of 16 band-pass filters that splits it into 16 frequency bands, from low to high, like a graphic equalizer. For each band an envelope follower measures, moment by moment, how much energy there is in that band. The carrier passes through an identical bank of 16 filters; the output of each filter goes through an amplifier (VCA/DCA) whose volume is controlled by the envelope of the corresponding modulator band. So when the voice pronounces an “a” with energy at 700 Hz and 1.2 kHz, exactly those synth bands open: the synth says the vowel.

Why you need a rich carrier

The vocoder does not add harmonics: it can only let through those the carrier already has. A sawtooth or a square wave (or noise) contains energy in all the bands and makes words clear; a sine contains only one and the word disappears.

Sibilant consonants

Fricatives (s, f, z, sh) have energy above 4–5 kHz, often outside the carrier: for this reason vocoders inject noise (hiss) or pass the modulator's high frequencies straight through when they detect them, improving intelligibility.

Pitch

The carrier decides the pitch: the notes you play on the keyboard (or, with pitch tracking, those detected from the voice) become the notes of the vocoded voice. By playing chords you get a harmonized choir from a single voice.

Attack, release and Shift

The times of the envelope followers decide how fast the vocoder reacts (crisp or soft articulation); shifting the frequencies of the carrier bands relative to those of the modulator (Shift) changes the formant and therefore the character of the voice (darker or higher).

Character and tone

The synthetic, metallic, robotic voice par excellence: the word stays understandable but takes on the timbre of the synth carrying it. With 16 bands it has the classic resolution of high-end analog vocoders: enough for intelligibility, with a soft, warm character compared with digital many-band vocoders (more surgical but less musical). The internal 2-VCO synth with wavefolding, FM and Ensemble gives bright carriers (saw and square) ideal for making words understood. Hiss/Buzz improve the sibilants. Strengths: wide choirs, robot voices, sung leads with pitch tracking, rhythmic articulation of pads through drums. Limitations: it requires a clean, well-leveled modulator signal, latency and routing in the DAW need care, and words lose intelligibility with carriers poor in harmonics (sine, triangle) or with too many bands attenuated.

Architecture

Original

• The vocoder was born at Bell Labs (Homer Dudley, 1930s) to compress the telephone band; it was used in cryptography during World War II.
• Robert Moog built a solid-state vocoder in 1968; in 1978–79 Moog Music produced the 16 Channel Vocoder (designed with Harald Bode), with two banks of 16 filters, envelope followers and VCAs, separate inputs for modulator and carrier, sibilance control and a matrix for rerouting the bands; it is the declared reference for Vocoder V (Bob Moog's 16-band voice box from 1979).
• The original had no internal synth: it required an external synthesizer as the carrier.

Vocoder V (Arturia, 2020)

• Virtual-analog modeling (TAE) of the 16-band banks with Attack (1 ms–1 s), Release (10 ms–30 s), Shift and Width of the bands, 16 per-band level faders with meters, High Freq section (Switched/Direct, Level, Hiss/Buzz Balance, Fast/Slow), Sample and Hold (freeze), modulator→carrier patch bay with presets, dry/vocoded Mix and Master Volume.
• Voice input with Gain, Threshold (generates note on/off), pitch tracking with Quantize, input Param EQ and compressor.
• Sample Player with 12 cells (WAV, AIFF, FLAC; over 700 factory samples), Keyboard modes (lowest octave reserved for triggers) and Cycle (forward, backward, forward/backward, random), time-stretch in multiples or bars, start/end, loop, Legato.
• Carrier synth: 2 VCOs (triangle, sawtooth, square, white noise) with Shape (PWM, noise color or wavefolding), Freq ±60 semitones, Level; Sync on VCO 1, FM from VCO 2 to VCO 1, VCO 2 Fixed and Fine; Bend Range, Glide up to 10 s, Hold, Chord (11 types plus Custom), Attack and Release 2 ms–20 s, Ensemble, Master Tune.
• Modulation: Envelope Follower (Gain, Rise, Fall; two destinations), polyphonic LFO (6 waveforms, 0.001–200 Hz or sync with straight/dotted/triplet values, Single/Cycle, Retrig, Unipolar; two destinations), 6-slot Assign (Carrier Envelope, Velocity, Keyboard, Aftertouch, Mod Wheel) to 30 destinations; 4 macros (Brightness, Timbre, Time, Movement).
• 3 effect slots in series or parallel with 11 effects: Reverb, Delay, Chorus, Flanger, Phaser, Overdrive, Compressor, BitCrusher, Multi Filter, Param EQ, Stereo Pan.

The controls, one by one

Top bar

Voice / Sample Player

Selects the modulator source: real-time audio input or internal samples (not both).

Advanced

Shows the advanced panel with Voice, Mods and Effects.

Gear icon (side panel)

Settings, MIDI learn, macros and tutorials.

Voice — Voice Input

Help

Routing instructions for the supported DAWs (sidechain or plug-in audio input).

Gain

Input preamplification from −36 to +24 dB; target: peaks around −18 dBFS in the display.

Threshold

−70/+6 dB: above the threshold the input generates note-ons for the synth, below it note-offs. Set it after Gain to prevent background noise from making the carrier sound.

Pitch Track (on/off)

Detects the pitch of a monophonic signal (voice, guitar, wind instrument) and plays the corresponding note on the synth: you sing the synth part.

Quantize

Snaps the detected notes to the chromatic scale (more robotic effect, perfect intonation).

Param EQ and Compressor (input)

Parametric EQ before the vocoder (cut below 100 Hz, emphasize 2–5 kHz for the consonants) and a single-slider compressor with automatic make-up to level the voice.

Voice — Sample Player

12 cells

Load WAV, AIFF, FLAC by drag & drop or from the browser (Factory, User, +Add Sample, +Add Folder); Alt+drag copies a cell.

Keyboard mode

The lowest octave (from C1) triggers the 12 samples, the rest of the keyboard plays the carrier.

Cycle mode and Play Direction

Each new note or chord starts the next sample: forward, backward, forward/backward, random; Reset returns to the start (also MIDI Start).

Lock

Locks a cell in Cycle so it always plays.

Stretch (Multi or Bars)

Time-stretch of the sample from 0.125× to 4× or from 1/32 to 8 bars, synced to the tempo.

Start / End

Start and end points (isolate single words of a phrase).

Gain

−36/+24 dB to even out the samples.

Loop

Forward loop within the region.

Legato

In Cycle: with legato notes the sample continues instead of moving to the next one.

Carrier Synthesizer — VCO 1 and VCO 2

Waveform

Triangle, Sawtooth, Square, White noise. Saw and square give maximum intelligibility; noise produces whispered voices or crowds.

Shape

On square: pulse width; on noise: color; on triangle and saw: wavefolding that adds high harmonics (more brightness and clarity).

Freq

Transposition ±60 semitones: VCO 2 at +12 for higher voices, −12 for deeper voices.

Level

Oscillator level.

Sync (VCO 1)

Hard sync: moving the Freq of VCO 1 changes the timbre, while VCO 2 decides the pitch.

FM

VCO 2 frequency-modulates VCO 1: metallic, dissonant timbres, alien voices.

Fixed (VCO 2)

VCO 2 at fixed pitch independent of the keyboard (drone or constant noise).

Fine (VCO 2)

Fine tuning in cents: a few cents of difference thicken the carrier.

Carrier Synthesizer — Globals

Bend Range

±1–24 semitones.

Glide

Portamento up to 10 s, also on chords (rises or falls depending on the first note played).

Hold

Holds a note or chord indefinitely (with Voice Input) or for the length of the sample: you play the chord once and sing over it.

Sel and Chord

Selects one of the 11 chord types or a Custom chord; with Chord on, one note plays the whole chord.

Attack

Carrier fade-in 2 ms–20 s.

Release

Carrier tail 2 ms–20 s.

Ensemble

Ensemble effect that thickens and widens the stereo image: essential for choirs.

Master Tune

400–480 Hz.

Keyboard and bottom bar

Virtual keyboard and wheels

Velocity depends on the click position; Bend with return to center; assignable Mod wheel.

Polyphony selection

Carrier polyphony (mono for leads and monotone robot, poly for choirs).

Undo, Redo, History, CPU/Panic

Edit management and note reset.

Macro 1–4

Brightness, Timbre, Time, Movement (configurable in the side panel with limits for each control).

Vocoder — Input, Times, Bands

Voice level and Synth level

Modulator and carrier levels into the vocoder (−75/0 dB); avoid the red light.

Preview (headphones)

Listen to the modulator alone or the carrier alone (diagnostics, not saved). The carrier in preview is played like a normal synth.

Attack

1 ms–1 s: response time of the followers. 2–10 ms = crisp articulation; slower = soft, fewer transients.

Release

10 ms–30 s: how long the carrier stays open after the modulator falls silent. 40–120 ms for clear speech; long for pads and freeze.

Shift

Shifts the band frequencies: to the left a darker, bigger voice, to the right a higher, smaller one.

Width

Width of the 16 filters: narrow = more synthetic and robotic; wide = more natural and full.

16 band faders with meters

Output level of each carrier band: lower the bands that clash with other instruments, raise 2–5 kHz for intelligibility; you can draw them by dragging.

Vocoder — High Freq, Master, Patch Bay, S&H

Switched / Direct

Switched: adds high frequencies only when the sibilants are strong (more robotic); Direct: always (more natural).

Level (High Freq)

Amount of high frequencies injected: set it to balance vowels and consonants.

Balance (Hiss/Buzz)

Mix between white noise (Hiss) and generated buzz (Buzz) for the sibilants; center positions are more natural.

Fast / Slow

The sibilants arrive as percussive bursts (Fast) or softer (Slow).

Master Volume

Output −60/0 dB.

Mix

Balance between dry and vocoded signal: a bit of dry voice increases intelligibility.

Patch Bay

Connects one modulator band (top row) to up to three carrier bands (bottom row); each carrier band receives only one signal. Inversions and shifts create alien voices. Bypass on/off and presets (Blank to clear).

Sample / Hold

Freezes the band levels: the carrier keeps the captured vowel or timbre while you play new notes.

Mods — Envelope Follower

On/Off

Enables modulation from the envelope of the input signal (after EQ and compressor).

Gain

−60/+30 dB to adapt the amplitude to the control range.

Rise

0.001–5 s: slows down the rise of the envelope.

Fall

0.001–5 s: slows down the fall.

Destinations 1 and 2 with Amount

Two destinations with amount ±1 (e.g. more voice = more Shape or more Ensemble).

Mods — LFO

Waveform

Sine, Triangle, Sawtooth, Ramp, Square, SnH.

Rate and type

0.001–200 Hz or rhythmic values (8 bars–1/32), straight, dotted or triplet.

Single / Cycle

A single cycle per note (envelope) or continuous cycle.

Retrig

In Cycle, restarts the LFO on every note.

Unipolar

Positive modulation only.

Destinations 1 and 2 with Amount

Polyphonic LFO (one per voice) to two destinations (e.g. slow Shift for breathing voices).

Mods — Assign (6 slots)

Source

Carrier Envelope, Velocity, Keyboard (bipolar around C3), Aftertouch (channel or polyphonic in standalone), Mod Wheel.

Destination and Amount

One of 30 destinations (VCO parameters, synth globals, vocoder parameters such as Shift, Width, Attack, Release, High Freq…) with amount ±1.

Effects (3 slots)

Series / Parallel

Post-vocoder effects in a chain or in parallel; each slot with bypass, selection and Wet/Dry.

Reverb

From room to hall.

Delay

Time, feedback, tone, stereo width, ping-pong, sync.

Chorus

Multiplies and widens the voice.

Flanger

Metallic effect and sweep, with sync.

Phaser

Phase sweep.

Overdrive

Solid-state saturation with tone.

Compressor

Peak control and tightness.

BitCrusher

From 16 to 1.5 bits and sample rate reduction: broken-radio or video game voices.

Multi Filter

Multimode filter with various slopes and resonance.

Param EQ

5 bands: 3 parametric plus low and high shelving.

Stereo Pan

Autopan.

Sound design tips

The modulator matters most

• Record or send a dry voice, close to the microphone, without reverb: reverb dirties the followers and the words blur together.
• Compress the input voice (internal compressor at halfway or more) and work at a constant level: a vocoder with uneven dynamics sounds patchy.
• Cut below 80–120 Hz in the input Param EQ (pops, rumble) and, if the words are unclear, boost the 2–5 kHz area by 3–6 dB.
• Set Threshold just above the noise floor: otherwise the carrier sounds even when you aren't speaking.

The carrier must be rich

• Sawtooth or square, possibly with Shape (wavefolding) and a touch of noise: more harmonics = more intelligibility. Avoid pure triangles as the only source.
• Two VCOs at +0 and +12, or slightly detuned with Fine and Ensemble, give wide choirs; a single VCO in mono gives the classic robot.
• Play chords in the middle range (C3–C5) for natural choirs; voicings that are too low muddy the low bands.

Bands and frequencies

• Vowels live between 300 Hz and 3 kHz, consonants between 2 and 8 kHz: don't lower the high bands too much if you want the words to be understood. Bands below 150 Hz can be attenuated to leave room for kick and bass.
• Use the band faders as a creative EQ before the effects: it is more musical than equalizing afterward.
• Narrow Width and Shift to the right = small, synthetic voice; wide Width and Shift to the left = big, dark voice.

Times

• Attack 2–10 ms and Release 40–120 ms for clear speech; long Releases (300 ms–1 s) for choir pads; too short gives a grainy sound.
• High Freq Switched for robotic voices, Direct for natural voices; set Level carefully to avoid disproportionate sibilants.

Non-vocal uses

• Drums as the modulator and a held chord (Hold) as the carrier: the pad moves in rhythm; lower the low bands if the kick opens the carrier too much.

Common mistakes

• Forgetting the routing in the DAW: with no modulator signal the vocoder stays silent.
• Monitoring with high latency while recording live: reduce the buffer.
• Mix all the way to vocoded with a barely intelligible voice: add 10–20% of dry signal.
• Using Sample Player and Voice Input together: not possible, choose one source.

Presets to recreate

In genres and records

• Electronic and synthpop: the Moog 16-channel vocoder is among the instruments used by Yellow Magic Orchestra, Devo and Saga; Kraftwerk make the vocoder a central element of their music, for example with the synthetic voices of The Robots, and on Autobahn some vocal parts are run through the vocoder.
• Funk, soul and jazz-funk: Sly and the Family Stone already used it in 1969 on Sex Machine; Stevie Wonder on Send One Your Love and A Seed's a Star; Herbie Hancock in the second half of the '70s; Midnight Star on Freak-A-Zoid.
• Pop and rock: Neil Young uses a Sennheiser vocoder on six tracks of Trans (1982); Phil Collins a Roland VP-330 on In the Air Tonight; Laurie Anderson a VP-330 on O Superman.
• French house and dance: Daft Punk use it from their first album to their last, from Around the World (vocals entirely run through the vocoder) to Get Lucky and Instant Crush.
• Film scores and TV: Wendy Carlos and John Carpenter are among the users of the Moog vocoder; the Cylon voices of Battlestar Galactica come from an EMS Vocoder 2000, and part of the melody of the 1980 Doctor Who theme from a Roland SVC-350.
• Electro and old-school hip hop: robotic voices and choirs.
• Nu disco, synthwave, electro-pop and indietronica: harmonized choirs and sung leads.
• Hyperpop and experimental pop: transformed and harmonized voices.
• Detroit techno and electro; video game and radio music.