Skip to main content

Plaits engines

This list was adapted from the original Plaits manual, with some edits to match Tidal's parameter implementation.

The engine parameter (0-15) can be set to select one of the models listed below.

All engines accept the harm, timbre and morph parameters, which have specific ways to shape the sound in each engine. The original Plaits module has an additional AUX output which features a distinct rendering of the original sound; in Tidal, you can set the mode parameter to 1 to get the equivalent to the AUX output.

0: Pair of classic waveforms

Virtual-analog synthesis of classic waveforms.

harmdetuning between the two waves
timbrevariable square, from narrow pulse to full square to hardsync formants
morphvariable saw, from triangle to saw with an increasingly wide notch (Braids’ CSAW)
mode 1sum of two hardsync’ed waveforms, the shape of which is controlled by morph and detuning by harm

A narrow pulse or wide notch results in silence! Use this trick if you want to silence one of the two oscillators, to get a variable square or variable saw.

1: Waveshaping oscillator

An asymmetric triangle processed by a waveshaper and a wavefolder. Sounds familiar? That’s the same signal processing chain as in Tides, when it runs at audio rate!

harmwaveshaper waveform
timbrewavefolder amount
morphwaveform asymmetry
mode 1variant employing another wavefolder curve, as available in Warps

2: Two operator FM

Two sine-wave oscillators modulating each other’s phase.

harmfrequency ratio
timbremodulation index
morphfeedback, in the form of operator 2 modulating its own phase (past 0.5, rough!) or operator 1’s phase (before 0.5, chaotic!)
mode 1sub-oscillator

Note: Set morph to 0 to get the same range of sounds as Braids’ WTFM. Set it to 1 to recreate the same sounds as Braids’ FBFM. A gentler palette equivalent to Braids’ FM is found with morph at 0.5.

3: Granular formant oscillator

Simulation of formants and filtered waveforms through the multiplication, addition and synchronization of segments of sine waves.

harmfrequency ratio between formant 1 and 2
timbreformant frequency
morphformant width and shape; this controls the shape of the window by which a sum of two synchronized sine oscillators is multiplied
mode 1simulation of filtered waveforms by windowed sine waves – a recreation of Braids’ Z*** models. harm controls the filter type (peaking, LP, BP, HP), with smooth variation from one response to another

4: Harmonic oscillator

An additive mixture of harmonically-related sine waves.

harmnumber of bumps in the spectrum; starts with one big bump, and progressively adds ripples around it
timbreindex of the most prominent harmonic; this control is somewhat similar to the cutoff frequency of a band-pass filter
morphbump shape – from flat and wide to peaked and narrow; this control is somewhat similar to the resonance of a band-pass filter
mode 1variant including only the subset of harmonics present in the drawbars of a Hammond organ (frequency ratios of 1, 2, 3, 4, 6, 8, 10 and 12)

5: Wavetable oscillator

Four banks of 8x8 waveforms, accessed by row and column, with or without interpolation.

harmsets the active bank (read below)
timbrerow index; within a row, the waves are sorted by spectral brightness (except for bank D which is a mess!)
morphcolumn index
mode 1low-fi (5-bit) output

There are 4 interpolated banks followed by the same 4 banks, in reverse order, without interpolation.

  • Bank A: harmonically poor waveforms obtained by additive synthesis (sine harmonics, drawbar organ waveforms).
  • Bank B: harmonically rich waveforms obtained by formant synthesis or waveshaping.
  • Bank C: wavetables from the Shruthi-1 / Ambika, sampled from classic wavetable or ROM playback synths.
  • Bank D: a joyous semi-random permutation of waveforms from the other 3 banks.

(TODO: make it clearer which values of harm select each bank, I didn't test)

6: Chords

Four-note chords, played by virtual analogue or wavetable oscillators. The virtual analogue oscillators emulate the stack of harmonically-related square or sawtooth waveforms generated by vintage string&organ machines.

harmchord type
timbrechord inversion and transposition
morphwaveform; values until 0.5 go through a selection of string-machine like raw waveforms (different combinations of the organ and string “drawbars”), and above 0.5 it scans a small wavetable containing 16 waveforms
mode 1root note of the chord

The proper values for harm (chord type) are

0.00 - 0.08octave
0.09 - 0.175
0.18 - 0.26sus4
0.27 - 0.36m
0.37 - 0.46m7
0.47 - 0.56m9
0.57 - 0.66m11
0.67 - 0.75M 6/9
0.76 - 0.85M9
0.86 - 0.95M7
0.96 - 1M

7: Vowel and speech synthesis

A collection of speech synthesis algorithms.

harmcrossfades between formant filtering, SAM, and LPC vowels, then goes through several banks of LPC words
timbrespecies selection, from Daleks to chipmunks. How does it work? This parameter either shifts the formants up or down independently of the pitch; or underclocks/overclocks the emulated LPC chip (with appropriate compensation to keep the pitch unchanged)
morphphoneme or word segment selection. When harm is greater than (0.4? original docs say knob at 11o'clock), a list of words can be scanned through
mode 1unfiltered vocal cords’ signal

8: Granular cloud

A swarm of 8 enveloped sawtooth waves.

harmamount of pitch randomization
timbregrain density
morphgrain duration and overlap; when set to 1, the grains merge into each other: the result is a stack of eight randomly frequency-modulated waveforms
mode 1variant with sine wave oscillators

To get a nice “supersaw” waveform, try a moderate amount of pitch randomization and grain density, with full grain overlap.

9: Filtered noise

Variable-clock white noise processed by a resonant filter. The cutoff frequency of the filter is controlled by freq. This allows proper tracking!

harmfilter response, from LP to BP to HP
timbreclock frequency
morphfilter resonance
mode 1variant employing two band-pass filters, with their separation controlled by harm

10: Particle noise

Dust noise processed by networks of all-pass or band-pass filters.

harmamount of frequency randomization
timbreparticle density
morphfilter type – reverberating all-pass network before 0.5, then increasingly resonant band-pass filters
mode 1raw dust noise

11: Inharmonic string modeling

No info on the original docs

12: Modal resonator

For your own pleasure, a mini-Rings! Refer to the Rings section for more information about modulated/inharmonic string synthesis, and modal resonators.

When the TRIG input is not patched, the string/resonator is excited by dust (particle) noise. Otherwise, the string is excited by a short burst of filtered white noise, or by a low-pass filtered click. (FIXME: what does the TRIG input equate to in Tidal?)

harmamount of inharmonicity, or material selection
timbreexcitation brightness and dust density
morphdecay time (energy absorption)
mode 1raw exciter signal

Note that Plaits uses a less powerful processor than Rings, and is thus limited to 3 voices of polyphony in inharmonic string modeling mode, and 1 voice of polyphony with 24 partials in modal resonator mode. Plaits does not allow you to control the position of the excitation, which is set to 25% of the length of the string/bar/tube.

13: Analog bass drum model

No fancy acronyms or patented technology here… Just behavioral simulation of circuits from classic drum machines! The drum machine employs a bridged T-network excited by a nicely shaped pulse.

harmattack sharpness and amount of overdrive
morphdecay time
mode 1frequency-modulated triangle VCO, turned into a sine with a pair of diodes, and shaped by a dirty VCA

Without any signal patched to the TRIG input, a continuous tone is produced. Not particularly useful, but its amplitude can still be modulated by morph and CV input (FIXME: equivalent to CV input in Tidal?).

14: Analog snare drum model

The generator employs a bunch of bridged T-networks, one for each mode of the shell, excited by a nicely shaped pulse; plus some band-pass filtered noise.

harmbalance of the harmonic and noisy components
timbrebalance between the different modes of the drum
morphdecay time
mode 1a pair of frequency-modulated sine VCO, mixed with high-pass filtered noise

15: Analog hi-hat model

A bunch of square oscillators generate a harsh, metallic tone. The resulting signal is mixed with clocked noise, sent to a HPF, then to a VCA. It uses 6 square oscillators and a dirty transistor VCA.

harmbalance of the metallic and filtered noise
timbrehigh-pass filter cutoff
morphdecay time
mode 1three pairs of square oscillators ring-modulating each other, and a clean, linear VCA