Overview

Chant is a formant synth. It makes vowels the way IRCAM's CHANT program did in the 1980s: every pitch period it fires a handful of tiny grains, one per formant, and together they sound like a voice. The key you play sets the pitch. The vowel pad sets what the voice is singing.

Drag across the pad and the sound slides from ee to ah to oo without the pitch moving. Pick a Bass, Tenor, Alto or Soprano formant table to match the register you're playing. Add a bit of vibrato, jitter and breath and you have a choir. Shorten the grain decay and the same engine gives you bells, plucks and overtone singing.

There's also a second grain model, VOSIM, which builds each period from a burst of pulses. It's buzzier and more electronic, closer to an old speech chip than a singer, and great for robots, reeds and talking basses.

Credits

Xavier Rodet, "Time-Domain Formant-Wave-Function Synthesis", Computer Music Journal 8(3), 1984. The FOF grain itself.
Xavier Rodet, Yves Potard and Jean-Baptiste Barrière, "The CHANT Project: From the Synthesis of the Singing Voice to Synthesis in General", Computer Music Journal 8(3), 1984. The parallel-formant voice, vibrato and jitter.
Werner Kaegi and Stan Tempelaars, "VOSIM: A New Sound Synthesis System", Journal of the Audio Engineering Society 26(6), 1978.
Michael Clarke, "FOF and FOG synthesis in Csound", in R. Boulanger (ed.), The Csound Book, MIT Press, 2000.
The Csound manual: the fof and vosim opcodes, and the formant tables Chant's Bass, Tenor, Alto and Soprano voices come from.

Features

Two grain models: smooth, vocal FOF and buzzy, reedy VOSIM.
Five formants (F1 to F5) running in parallel, like the human vocal tract.
F1 × F2 vowel pad with the vowels i, e, a, o, u marked. Everything in between blends smoothly.
Bass, Tenor, Alto and Soprano formant tables.
Per-formant offsets in Hz and dB, for a brighter singer's formant, a whistling F2 or inharmonic bells.
Grain controls: rise, decay, length and octaviation, which drops the pitch an octave without moving the vowel.
Vibrato, jitter, breath and glide on every voice. Each voice starts its vibrato at a different phase, so chords sound like a section, not one singer.
Live displays: a grain view, a spectrum showing harmonics under the formant envelope, and a ring on the vowel pad that follows modulation.
8-slot mod matrix with two envelopes, two tempo-syncable LFOs with delay, velocity, key track, mod wheel, pitch bend, aftertouch and random. Route anything to the vowel axes and the voice starts to talk.
Distortion, chorus, delay and reverb on the master bus.
100 factory presets: choirs, solo voices, drones, bells, plucks, leads, basses, keys, percussion, strings, brass and some very strange FX.
Copy and paste whole patches as text.