Polyphony has two meanings in music. In music theory it is a texture made of two or more independent melodic lines sounding at the same time, as in a Renaissance motet or a Bach fugue. In electronic instruments it is the number of notes an instrument can sound at once: an 8-voice synthesizer has 8-note polyphony, a digital piano might have 128.

The two meanings share a Greek root, polyphōnia (“many sounds” or “many voices”), but they answer different questions. The theory meaning describes how a piece is written; the technical meaning describes what a piece of gear can do. A synth with 16-voice polyphony can play a homophonic chord accompaniment, and a monophonic synth can take part in a polyphonic piece if each line is recorded on its own track.

Table of Contents

What is polyphony?

Polyphony as musical texture

Texture describes how the layers of a piece relate to each other. In a polyphonic texture each line is a melody in its own right, with its own rhythm and contour, and the lines combine to form harmony as they move. Encyclopaedia Britannica defines polyphony as “the simultaneous combination of two or more tones or melodic lines” and notes that a texture becomes more purely polyphonic, and more contrapuntal, the more the lines are rhythmically differentiated (Britannica). Merriam-Webster gives counterpoint as a synonym.

In plain terms: if a choir sings the same words in the same rhythm, moving together from chord to chord, you hear one block of sound. If the sopranos start a phrase, the altos enter with the same tune two beats later, and the tenors and basses follow with their own versions, you hear several conversations at once. The second case is polyphony.

Polyphony as an instrument specification

On a synthesizer, sampler, digital piano or software instrument, polyphony is the maximum number of voices the instrument can produce at the same moment. A voice is one complete sound-generating path: on an analog synth, oscillators, a filter, an amplifier and the envelopes that shape them; on a sampler, one playing sample with its own envelope and filter. An instrument that can play only one note at a time is monophonic; one that can play several is polyphonic, usually quoted with a number (4-voice, 8-voice, 16-voice).

A third meaning: ethnomusicology

Ethnomusicologists use polyphony more broadly for any music in which several pitches sound together in group singing, including drones, parallel intervals and overlapping call-and-response. Traditional vocal polyphony is central to many cultures; Georgian polyphonic singing, for example, was recognized by UNESCO as part of the world’s intangible cultural heritage. This broader use explains why some sources describe a drone under a melody as polyphonic while strict counterpoint teachers would not.

History of polyphony

From chant to organum

Early medieval church music was monophonic: plainchant, a single unaccompanied line. The first written descriptions of adding a second voice come from the anonymous treatise Musica enchiriadis of about 900, which describes organum: a second voice added to the chant at a fixed interval such as the fourth or fifth, largely moving in parallel with it.

Over the following centuries the added voice gained independence: it began to move in contrary motion, then to sing long, florid melismas over slowly sustained chant notes. In Paris around 1200 the Notre-Dame school took the decisive step. Léonin is associated with the Magnus liber organi, a large book of two-voice organa, and his successor Pérotin wrote organa in three and four parts. Pérotin’s four-voice Viderunt omnes and Sederunt principes are linked to performances at Notre-Dame in 1198 and 1199. These pieces also used rhythmic modes, the first systematic way of coordinating several rhythmically independent lines.

Thirds, sixths and the Renaissance

In the late Middle Ages English singers favored thirds and sixths, intervals older theory had treated as imperfect. On the Continent, fauxbourdon (harmonizing a melody in parallel sixth chords) appears in sources from around 1430, with Guillaume Du Fay among its earliest known users. The sound of full triads became normal, and Renaissance music brought polyphony to a peak: four, five or six equal voices, imitating one another, as in the masses and motets of Josquin des Prez and Giovanni Pierluigi da Palestrina.

Counterpoint as a discipline

In 1725 Johann Joseph Fux published Gradus ad Parnassum, a textbook that taught counterpoint step by step through five “species,” from note-against-note writing to florid lines, using Palestrina’s style as its model. It shaped composition teaching for two centuries. Johann Sebastian Bach (1685–1750) is the composer most often cited as the summit of polyphonic writing: The Well-Tempered Clavier (Book I compiled in 1722, Book II in the early 1740s) contains 48 preludes and fugues, and every fugue is a study in how independent lines can coexist. After Bach, homophonic textures dominated the Classical period, but fugue and counterpoint never left the composer’s toolkit.

Electronic polyphony before microprocessors

Early electronic instruments were monophonic because each oscillator was expensive and unstable. Harald Bode’s Warbo Formant-Orgel of 1937 is described as an early archetype of a polyphonic synthesizer with voice allocation. The Hammond Novachord of 1939, designed by Laurens Hammond, John M. Hanert and C. N. Williams, is widely described as the first commercial polyphonic synthesizer. It used 12 top-octave oscillators and frequency dividers to derive every note of its 72-note keyboard, all of which could sound at once. It used 163 vacuum tubes; about 1,069 units were built before production ended in 1942 (Wikipedia: Novachord).

The same divide-down principle powered electronic organs and the string machines of the 1970s. Moog’s Polymoog (1975) combined divide-down tone generation with per-key amplifier and filter circuits, giving 71-note polyphony, but had only one programmable filter shared by all notes, which limited its programmed sounds (Sound On Sound).

Voice allocation and the analog polysynth

True synthesizer polyphony needs a complete voice per note and a way to assign keys to voices. In the mid-1970s Dave Rossum (of E-mu Systems) developed a digitally scanned keyboard that did exactly that. Oberheim used it to link several of its Synthesizer Expander Modules: the Oberheim Four Voice (1975) contained four SEMs, one per note, and an Eight Voice soon followed. Yamaha’s CS-80 (1977) offered 8 voices, each with two layers, and a keyboard with velocity and polyphonic aftertouch, meaning pressure could be applied to individual notes rather than to all notes at once.

In 1978 Sequential Circuits released the Prophet-5, designed by Dave Smith and John Bowen: five voices, a microprocessor handling keyboard scanning and voice assignment, and memory that stored complete sounds. It is generally credited as the first polyphonic synthesizer with fully programmable memory, and it set the template for the instruments that followed.

MIDI, digital synthesis and software

MIDI 1.0 (1983) gave polyphonic instruments a shared language, with separate note-on and note-off messages for every key. The same year Yamaha’s DX7 brought 16-voice digital FM polyphony to a mass market. Digital synthesis and sampling made voices cheaper with every generation; today software instruments are limited mainly by the computer’s processing power, while analog flagships such as the Moog One (2018, in 8-voice and 16-voice versions) still treat voice count as a defining specification.

How polyphony works in instruments

What a voice is

On a classic subtractive synthesizer one voice contains one or more oscillators, a filter, an amplifier and at least two envelopes (one for the filter, one for the amplifier), typically ADSR envelopes. A polyphonic synth duplicates that whole path for every note it can play at once. When you press a chord, each note gets its own voice, so each note starts its attack when its key goes down and starts its release when its key comes up, independently of the others.

Voice allocation

A controller (a microprocessor in hardware, code in software) watches the keyboard or incoming MIDI and assigns each new note to a free voice. Assignment schemes vary: some reuse the voice the same key used last time, which helps analog synths sound consistent; others rotate through the voices in turn, which spreads notes across slightly different analog circuits and lets the release tails of previous notes ring on. On vintage polysynths small differences between voice cards are part of the sound, which is why software emulations often add per-voice variation.

Voice stealing

When every voice is busy and a new note arrives, the instrument must either ignore the new note or steal a voice that is already sounding. Two different ideas are often mixed up here: note priority (which notes the instrument protects) and the stealing rule (which sounding voice gets cut).

ModeWhat happens when voices run outTypical use
Last-note priority (steal oldest)The note that has been sounding longest is cut and its voice plays the new noteThe most common default; new notes always sound
First-note priorityNew notes are ignored until a key is releasedRare; protects held chords
Highest-note priorityLower notes are stolen first, so the top line survivesKeeps a melody on top of chords intact
Lowest-note priorityHigher notes are stolen first, so the bass survivesProtects a bass line; also common on monophonic synths
Release-first stealingA voice already in its release phase is taken before a held oneUsed by many instruments to make stealing less audible

Stealing is audible because the stolen voice’s envelope is cut or restarted mid-note. At worst this produces a click as the waveform jumps; at best a release tail simply vanishes early. Instruments soften this with a very short fade on the stolen voice, but nothing can make a missing note come back.

Monophonic, duophonic, paraphonic, polyphonic

ArchitectureNotes at onceFilter and amplifierBehavior
Monophonic1One pathOnly one note sounds; legato and glide behave naturally. See mono for the separate meaning of mono audio.
Duophonic2Often sharedTwo oscillators can be split across two keys, as on some 1970s synths
ParaphonicSeveral pitchesShared by all notesEach note has its own oscillator, but a new note retriggers the common envelope, so all notes swell and fade together
Divide-downEvery keyOften shared or fixedAll keys can sound; tone comes from divided top-octave oscillators, as in organs and string machines
Fully polyphonicSet by voice countOne per voiceEach note articulates independently

Paraphony is common on modern semi-modular instruments because it delivers chords without the cost of a full voice per note. Moog’s Matriarch (2019), for example, offers mono, duo and 4-note paraphonic modes from four oscillators that share its filters and envelopes. Paraphonic playing works best for chords that start and stop together.

Unison and stacking

Unison mode assigns several voices to each key and detunes them slightly for a thicker sound. It uses up polyphony: an 8-voice synth in 4-voice unison can play only two notes. Software synths make the same trade in CPU rather than hardware. (In music theory unison means two parts on the same pitch, a related but different idea.)

Multitimbral instruments and samplers

A multitimbral instrument plays several different sounds at once, usually on different MIDI channels, and they share one pool of voices. Layered sounds multiply voice use: a 4-note chord on a patch that layers piano, strings, pad and choir needs 16 voices. On samplers and digital pianos one note can use more than one voice (stereo samples, layered velocity zones, release samples, pedal resonance), which is why manufacturers quote figures like 128 or 256 notes even though no pianist plays that many keys at once. Holding the sustain pedal during fast passages is the classic way to reach the limit.

MPE: polyphonic expression

In standard MIDI, pitch bend, channel pressure and most controllers apply to a whole channel, so bending one note of a chord bends them all. MIDI Polyphonic Expression (MPE) gives each sounding note its own MIDI channel, with a master channel for messages that affect everything. Each note can then carry its own pitch bend, pressure and a third dimension of movement, usually mapped to timbre. MPE became part of the official MIDI specification in 2018 (Sound On Sound). The MIDI Association later adopted an MPE Profile for MIDI-CI, in which the receiver reports the channels it supports and the sender adapts (MIDI Association). Polyphonic aftertouch, on instruments like the CS-80, was the earlier hardware form of the same idea.

Polyphony in practice

Writing polyphonic music

Counterpoint in the classical sense is about independence and balance. Composers and arrangers keep lines distinct through contrary and oblique motion (lines moving in opposite directions, or one holding while the other moves), different rhythms in each part, staggered entries and imitation. Parallel fifths and octaves are traditionally avoided because they make two lines merge into one. The same habits help in production: a bass line that moves against the top melody, a counter-melody that fills the gaps in a vocal, or a synth arpeggio that moves when the chords hold all create a polyphonic feel inside a mostly homophonic song.

Hearing the difference

  • Homophony: one clear tune, everything else moves with it as chords. Most pop songs, hymns and singer-songwriter tracks.
  • Polyphony: you can follow more than one tune, and the parts often start at different times. Fugues, rounds such as “Row, Row, Row Your Boat,” Renaissance motets, New Orleans–style collective improvisation in jazz and some progressive rock.
  • Heterophony: several performers play versions of the same melody with different ornaments at the same time, common in many Asian and Middle Eastern ensemble traditions.

How many voices do you need?

The voice count you need depends on the longest overlap between notes, not on the size of your chords. A sound with a long release keeps sounding after the key is lifted, so moving from one 4-note chord to the next needs up to 8 voices if the first chord is still ringing. Rough guidelines:

VoicesWhat fitsLimits
1 (mono)Bass lines, leads, acid lines, glide effectsNo chords
4Simple chords and stabs with short releasesPads with long releases steal voices on every chord change
6 to 8Two-handed playing, pads with moderate releasesLong releases and sustain pedal can still run out
16 and moreSustain-pedal playing, dense voicings, layered and multitimbral setupsCPU or cost, not musical range

Working around low polyphony

  1. Shorten the release in the patch and let a reverb or delay supply the tail. Effects run after the voices, so they add length without using any.
  2. Record in passes. Play the top notes of a chord on one take and the lower notes on another, or record each line of a counterpoint separately. A monophonic synth can build a full chord this way.
  3. Change the priority mode if the synth allows it, for example to protect the lowest note under a moving top line.
  4. Turn off unison or reduce the number of stacked voices when you need more notes.
  5. Freeze or bounce software instruments that hit CPU limits, then lower the voice limit in the plugin to keep projects stable.

Mono basses and polyphonic pads

Producers in house, techno, trap and other electronic genres often set bass synths to monophonic mode deliberately. Overlapping low notes can blur the low end and stack up level, and mono mode gives clean legato and glide between notes, which matters for 808-style slides. Pads, strings and keys are usually set to high polyphony with long releases, which is exactly where voice limits show up first.

Polyphony settings in DAWs and plugins

Most software instruments expose polyphony as a voices or polyphony setting, often next to mono, legato and glide options. Ableton’s Drift (introduced in Live 11.3) has Poly, Mono, Stereo and Unison voice modes, polyphony of up to 32 voices, and a Drift control that adds a different random variation to each voice’s oscillators and filter, imitating the voice-to-voice differences of analog polysynths (Ableton manual). Logic Pro’s Alchemy, Cubase’s instruments, Bitwig and most third-party synths offer similar controls, and Cubase’s VST3 Note Expression and Bitwig’s per-note editing let you draw expression for individual notes inside a chord. Samplers such as Kontakt let you set a maximum voice count per instrument, which is useful for keeping a project within CPU limits.

Instruments and tools

InstrumentYearPolyphonyWhy it matters
Hammond Novachord193972 notes (divide-down)First commercial polyphonic synthesizer
Moog Polymoog197571 notes (divide-down)Per-key articulation, one shared programmable filter
Oberheim Four Voice19754 voicesFour separate synth modules under one scanned keyboard
Yamaha CS-8019778 voicesPolyphonic aftertouch and per-note expression
Sequential Prophet-519785 voicesWidely credited as the first fully programmable polysynth
Yamaha DX7198316 voicesAffordable digital FM polyphony with MIDI
Moog One20188 or 16 voicesModern multitimbral analog flagship
Moog Matriarch20194-note paraphonicChords from shared filters and envelopes
Ableton Drift2023Up to 32 voicesSoftware polysynth with per-voice variation

Roland’s Linear Arithmetic synthesis instruments of the late 1980s are a useful example of how digital designs traded voices for complexity: a sound built from several partials uses several of the instrument’s voice resources, so richer patches mean fewer notes. The same trade applies to layered patches in any workstation.

Famous and documented examples

Polyphonic music

  • Pérotin, Viderunt omnes (linked to Notre-Dame, 1198): one of the earliest surviving four-voice compositions.
  • Thomas Tallis, Spem in alium (c. 1570): a motet for 40 independent voices in eight five-voice choirs.
  • Palestrina, Missa Papae Marcelli: the standard teaching model of Renaissance polyphony.
  • J. S. Bach, The Well-Tempered Clavier and The Art of Fugue: the reference works of Baroque counterpoint.
  • Mozart, finale of the “Jupiter” Symphony No. 41: a famous Classical-era combination of five themes in counterpoint.

Polyphonic synthesizers on record

  • Vangelis, Blade Runner (1982): the Yamaha CS-80 was Vangelis’s main synthesizer, and its expressive polyphonic sound features prominently on the score.
  • Van Halen, “Jump” (released December 1983): the keyboard part was played on an Oberheim OB-Xa.
  • Prince, “1999” (1982): credits list an Oberheim OB-Xa and an ARP Omni-2.
  • Gary Numan, “Cars” (1979): its sound is closely associated with the Polymoog’s presets.
  • Phil Collins, “In the Air Tonight” (1981): one of many hits that used the Prophet-5.

Common mistakes and myths

Myth: “A song with vocals and chords is polyphonic”

A melody over chords is homophony, even though many notes sound at once. Polyphony in the textural sense requires more than one melody. Calling pop songs polyphonic is a common confusion between the theory meaning and the everyday sense of “many notes.”

Myth: “Paraphonic is the same as polyphonic”

Both play several pitches at once, but a paraphonic synth runs them through shared filters and envelopes. Play a new note while holding a chord and every note responds to the retriggered envelope. On a polyphonic synth each note keeps its own envelope.

Myth: “Unison adds voices”

Unison stacks existing voices on each note. It makes each note bigger and reduces the number of notes you can play.

Mistake: counting chord notes, not overlaps

A 4-voice synth can play a 4-note chord, but not a sequence of 4-note chords with long releases without stealing voices. Count the notes that ring at the same moment, including release tails and sustain-pedal notes.

Mistake: blaming the synth for clicks

Clicks on chord changes are often voice stealing, not a fault. Shorten releases, lower the number of layers, or increase the voice limit in a plugin before suspecting your audio interface.

Confusing sampler round-robin with voice rotation

In samplers, round-robin means alternating between several recordings of the same note so repeated notes sound natural. It is a sample-selection feature, not a voice-allocation mode, although the name is the same.

TermMeaningDifference from polyphony
MonophonyA single melodic line, or an instrument that plays one note at a timeOne line or one voice only
HomophonyOne melody supported by chords moving with itMany notes, but one leading line
HeterophonySimultaneous variations of the same melodyOne melody, several versions
CounterpointThe technique and rules of combining independent linesCounterpoint is the craft; polyphony is the resulting texture
ParaphonySeveral pitches through shared filter and envelopeNo independent articulation per note
MultitimbralitySeveral different sounds at onceAbout number of sounds, not number of notes
PolyrhythmTwo or more conflicting rhythms at onceAbout rhythm, not melodic lines

Standards and research

  • MIDI 1.0 (1983): separate note-on and note-off messages per key make polyphonic sequencing possible; controllers apply per channel.
  • MPE (2018): per-note channels for polyphonic expression; extended by the MIDI-CI MPE Profile under MIDI 2.0.
  • Fux, Gradus ad Parnassum (1725): the historical reference for teaching species counterpoint.
  • Automatic transcription research: extracting the individual notes from a polyphonic recording is a long-standing problem in music information retrieval. Anssi Klapuri’s paper “Multiple Fundamental Frequency Estimation by Summing Harmonic Amplitudes” (ISMIR 2006) is a widely cited paper that proposes an iterative method (estimate the strongest pitch, cancel it, repeat) alongside a joint estimator (paper). Polyphonic pitch detection of this kind underlies audio-to-MIDI conversion and note-editing tools.
  • Melody: the building block of every polyphonic texture.
  • Harmony: the vertical result of lines sounding together.
  • Chord progression: the homophonic counterpart to contrapuntal writing.
  • Scale: the pitch material polyphonic lines are drawn from.
  • Renaissance music: the golden age of vocal polyphony.
  • ADSR: the envelope each synth voice carries, and the reason release tails use up voices.
  • MIDI: how polyphonic notes are sent between instruments.
  • Mono: single-channel audio, not to be confused with monophonic instruments.
  • Sound design: where voice count, unison and priority modes become creative choices.
  • Linear Arithmetic synthesis: a digital method where patch complexity affects available voices.
  • Arpeggiator: turns held chords into lines, a way to imply harmony on a mono synth.

FAQ

What is polyphony in simple terms?

In music theory, polyphony is two or more independent melodies played or sung at the same time. On a synthesizer or keyboard, polyphony is the number of notes it can sound at once.

What is the difference between polyphony and homophony?

Polyphony has several independent melodic lines. Homophony has one main melody with chords that move along with it, which describes most pop songs.

How many voices of polyphony do I need on a synth?

For two-handed playing and pads, 6 to 8 voices is a common minimum. Long releases and sustain-pedal playing need more; mono is enough for basses and leads.

What is the difference between polyphonic and paraphonic?

A polyphonic synth gives every note its own filter and envelopes. A paraphonic synth plays several pitches but sends them through a shared filter and envelope, so a new note retriggers the sound of the notes already held.

What is voice stealing?

When all voices are in use and a new note arrives, the instrument cuts one sounding note to free its voice. The rule for which note is cut is set by the priority or allocation mode.

Why do my synth chords click when I change chords?

Usually because the synth runs out of voices and steals ones that are still in their release. Shorten the release, reduce unison or layers, or use an instrument with more voices.

What does 128-note polyphony mean on a digital piano?

The piano can sound up to 128 voices at once. Stereo samples, layers and the sustain pedal can use several voices per key, so high figures matter more than they appear to.

What was the first polyphonic synthesizer?

The Hammond Novachord of 1939 is widely described as the first commercial polyphonic synthesizer. The Prophet-5 of 1978 is widely credited as the first polysynth with fully programmable memory.

Can you play chords on a monophonic synth?

Not live, but you can record each note of a chord on separate passes, or use an arpeggiator or delay to imply harmony.

How does MPE work?

MPE sends each note on its own MIDI channel, so pitch bend, pressure and timbre control can be applied to individual notes inside a chord.