mmmt-tools / Synthesis
on-device · nothing leaves this page

Synthesis

A synthesizer per method — each one playable from your computer keyboard, with a scope and spectrum so you can see what the technique does to the signal. Thirty-two instruments across six tiers.

The canon

The foundational methods every synthesist knows.

Subtractive

live

Rich oscillators carved by a resonant filter, with a filter envelope. The classic analog voice.

Additive

live

Build a tone by summing sine partials — including inharmonic bell ratios no single waveform can make.

FM

live

Two operators, a ratio, and an index — with live deviation, sideband and Carson-bandwidth readouts.

Wavetable

live

Scan through a table of single-cycle waveforms; band-limited, so sweeping never aliases.

Granular

live

Clouds of grains from a dropped sample — size, density, position, jitter, and the window that stops the clicks.

Sample Playback

live

Pitch a sample by playback rate, with a root note, loop points, and an amp envelope.

Modulation, distortion & shaping

Bend, multiply, and fold a signal into new harmonics.

AM & Ring

live

Multiply two oscillators into sum-and-difference sidebands — and hear why ring modulation drops the carrier and AM keeps it.

Waveshaping

live

Bend a sine through a transfer curve; Chebyshev polynomials let you dial in each harmonic directly, and drive changes the timbre.

Wavefolding

live

West-coast synthesis: fold a sine past its rails to add harmonics, with a low-pass gate that ties brightness to loudness.

Phase Distortion

live

The Casio CZ trick: warp the phase clock reading a sine and harmonics appear with no filter. A hand-written worklet oscillator.

Virtual Analog

live

Naive vs PolyBLEP oscillators, so you can hear aliasing — plus PWM and a unison supersaw. Worklet-generated.

Subharmonic

live

The undertone series: divide the played note by whole numbers to grow a just-intoned chord downward, Subharmonicon-style.

Physical modeling

Simulate the physics of strings, tubes, bars, and drum heads.

Karplus–Strong

live

A plucked string from a noise burst in a delay line — the simplest physical model, and a beautiful one. No oscillator anywhere.

Digital Waveguide

live

Two travelling-wave delay lines with reflecting ends: a plucked string with pick-position comb, or a breath-driven, self-oscillating tube.

Modal

live

Strike a bank of tuned resonators — bell, bar, drum — and the inharmonic partials no waveform can make ring and decay.

Membrane (FD)

live

A drum head from Newton's laws on a grid — the wave equation stepped sample by sample, with a stability limit you can hear.

Scanned

live

Read a slowly-vibrating mass-spring ring as a wavetable at audio rate, so a held note's timbre morphs on its own.

Spectral & analysis/resynthesis

Take the sound into the frequency domain, edit it, and bring it back.

Convolution

live

Smear a sound through an impulse response — a reverb is just decaying noise. IRs generated in code, or drop your own to cross-convolve.

Vocoder

live

A bank of band-pass filters and envelope followers: a voice's spectral shape drives a synth you play, so a sawtooth chord talks.

Spectral Freeze

live

STFT the sound, edit the bins, transform back — freeze a frame forever, blur it, gate a band, or flatten phase into a robot.

Frequency Shift

live

Add a fixed number of Hz to every partial — not multiply. The harmonic ratios break and the tone turns bell-like. The one people confuse with pitch.

Phase Vocoder

live

Separate time from pitch: stretch to twice the length at the same pitch, or shift an octave in the same length. Rendered offline.

Sinusoidal Modeling

live

Split a sound into stable sinusoids plus a noise residual, then rebalance them — a glassy de-noised tone or a breathy, pitchless ghost.

Voice & formant

Three ways to synthesize the human voice.

Formant / Vowel

live

The source–filter voice: a buzz through a bank of resonant formant filters. Morph between two vowels and hear a sung "aah → eee".

FOF Formants

live

The other way to make a voice: build each formant from a stream of grains fired once per vocal-fold period — CHANT-style granular synthesis.

LPC Talkbox

live

Fit an all-pole filter to a voice, keep the words, and replace the buzz with a pitch you play — the model behind every talking toy.

Stochastic, corpus & experimental

Statistics, chaos, and code as sound sources.

Vector Synthesis

live

Four sources, one joystick: crossfade between them with an XY pad and let the point travel on its own for an evolving pad.

Stochastic Clouds

live

Xenakis's idea: compose the statistics of a swarm of grains — set the distributions for arrival, pitch and loudness, not the notes.

Wave Terrain

live

Read a waveform off a 3-D surface by tracing an orbit across it — a small path change crosses different hills and reshapes the tone.

Chaotic Oscillators

live

The logistic map and its cousins: one knob sweeps from a pure tone through period-doubling into deterministic noise.

Bytebeat

live

Music from one line of integer arithmetic: bit-shift a counter, keep the low byte, and out comes melody and rhythm. Edit the formula live.

Concatenative

live

Play a recording by its sound: chop it into a corpus, then navigate a feature pad to grab and stitch the grains nearest your target.

Patchbay

Not a method — a small modular rig you wire yourself.

Modular Patchbay

live

Sixteen modules — oscillators, a step sequencer, sample & hold, ring mod, a suboctave, sends into delay and reverb — patched jack to jack, kind to kind, with the signal flow drawn as cables.

All sections