What is a “voice”, exactly?
A voice is one complete copy of the sound chain — oscillators, filter, amplifier — able to play one note. The word matters because keyboards ship with a fixed number of them, and the number decides what a chord means on that instrument.
Monophonic: one voice
A mono synth owns exactly one voice. Play a three-note chord and you get the last key pressed — the other two are silent. The idiom that grew out of this constraint treats chords as arpeggios: rolled note orders, not stacks. The mono synth chord pages chart every chord that way, numbering the roll low to high, because that is what the machine can actually play.
Paraphonic: many keys, one chain
A paraphonic keyboard tracks several pitches — four, on the paraphonic model charted here — but sends them all through one shared filter and envelope. Hold a chord and every note swells and fades together as a single block. It is a real, musical architecture (and historically a cheap one), but it cannot articulate inner voices: re-strike one key and the whole chord’s envelope restarts. The paraphonic chord pages name that trade-off on every chart.
The budget arithmetic
How often do voice limits actually bite? This site charts 62 chord qualities, from two-tone power chords up to six-tone 13ths. Exactly 34 of the 62 carry more than four tones — so on a 4-voice polysynth, more than half the catalog forces the reducer’s hand, while on eight voices every quality fits whole. That ratio is the whole argument for knowing your architecture: on small polys, chord voicing isn’t a style choice, it’s the operating condition.
Polyphonic: a voice per key, up to a budget
A polysynth carries a bank of complete voices — eight and four on the models here — and a voice allocator hands them out as you play. Within the budget, every note gets its own envelope and its own filter sweep. Past the budget, something has to give: that story is the voice-stealing guide. Compare the same chord on the 8-voice and 4-voice pages and watch the reducer make the call.