Kowane turns a voice into a handful of sine waves: sine-wave speech. Its companion, Kowane MIDI, turns the same analysis into notes, so a piano or any instrument can follow the voice and start to speak. The name comes from 声音 (kowane), the colour of a voice.

A voice made of sines

Kowane follows the voice and plays it back as 1 to 16 sine waves. With few sines the words are hard to follow; with more, the voice becomes clearer. Snap pulls the sines toward a scale, glide smooths their movement and width spreads them across the stereo field.

Kowane MIDI

Put Kowane MIDI in front of a piano or any instrument and pick the voice track in its from menu. Three modes turn the analysis into notes: formant, spectrum (the closest to speech) and envelope (vowel colour as clusters). Any number of Kowane MIDI can listen to one Kowane, each with its own settings.

Envelope

The envelope turns each step into a plucked bundle of sines with its own attack, decay, sustain and release, measured in steps or milliseconds, with up to eight overlapping layers.

Views

Three views show what Kowane hears: a vowel chart normalised to the speaker, a trace of the sine frequencies, and a mouth drawn in dots that opens and widens with the voice.

To sum it up

• Kowane (Audio Effect): 1–16 sine waves from a voice
• Kowane MIDI (MIDI Effect): formant, spectrum and envelope modes, up to 16 notes
• Snap toward 29 scales and chords, with glide, width, depth and shift
• Plucked envelope with up to 8 layers
• Voice / full source for speech or drums and instruments
• Vowel, trace and lips views
• Several Kowane MIDI can follow one Kowane, including on return tracks and in racks

Compatibility

• macOS — tested on Live 12.4 / Max 9.1.4
• Windows — not tested yet
• Ableton Live 12 Suite, or Live with Max for Live and Max 9

Most LFOs run on a clock: a rate in Hz or a note length. Hamon runs on water instead. Stones fall into a simulated water surface on their own, and five buoys float on the ripples; the height of each buoy becomes one of five LFOs. The movement is smooth, related across the five outputs, and never locked to the beat.

Water

Hamon simulates a single line of water. Nobody throws the stones: one falls at a random spot whenever the surface gets too calm, so the water keeps moving. Strength, Time and Smooth set how busy, how slow and how soft it is.

Mapping

Click Map, then click any parameter in Live. Each of the five rows sets its own min and max, so one LFO can sweep a filter wide open while another only nudges the pan.

Tempo

Hamon ignores Live's tempo and transport. Use it where tempo-synced LFOs sound mechanical: a filter on a pad, a wavetable position, reverb size, the pan of a texture. The scope view draws all five outputs over the last few seconds.

To sum it up

• Five LFO outputs, each mapped to any parameter in Live
• min / max range per output
• Strength, Time and Smooth for the water
• Center and Spread for the buoys (or drag on the display)
• Water view and LFO scope view
• Several Hamons run independently
• Runs whether or not Live is playing
• Frozen device: nothing else to install

Compatibility

• macOS — Apple Silicon / Intel
• Windows — 64-bit
• Ableton Live 12 with Max for Live (tested on Live 12.4)

Got Hamon on Gumroad before? Version 1.2 has been sent to the email address you used there. If it didn't arrive, write to shop@cotofu.com.

Video

This is not white noise. Glassy couples hundreds of copies of one simple chaotic map, x → 1 − a·x², and steps them at audio rate. Depending on two numbers, the nonlinearity a and the coupling epsilon, the elements move as one, split into a few synchronized groups, or scatter into turbulence. Each of these states has its own noise.

Listen

Rendered from the device's own engine and level-matched for listening (−16 LUFS). Glassy itself can be far louder; start with your volume down.

1. ▶ Between Phases 0:30 Drift pushes the coupling back and forth, so the sound keeps crossing between rough noise, buzzing tones and calmer stretches.
2. ▶ Two-Cluster Hum 0:20 The cells split into two groups swinging in opposite phase, and the noise turns into a steady buzz near 1.2 kHz.
3. ▶ Order Detector 0:20 The output is how far the cells disagree: it drops to silence whenever the lattice locks together and bursts back when it scatters.
4. ▶ Almost White 0:20 The fully chaotic ring gives the flattest spectrum of the set: the closest Glassy gets to white noise, and still not quite.

Kaneko's globally coupled map

In 1989 the physicist Kunihiko Kaneko coupled many copies of the logistic map so that each element feels the average of all the others. As the coupling weakens, the system passes through four phases: coherent, where every element moves as one; ordered, where it splits into a few clusters (at a = 1.8, two clusters swinging in opposite phase); partially ordered, where few-cluster and many-cluster states coexist and the lattice can wander between them for a long time; and turbulent, where the elements scatter, yet their average keeps fluctuating more than the law of large numbers allows. Physicists also call the partially ordered phase glassy. The word refers to glass in the physical sense — not the clear material, but a disordered state that freezes before it can find order. Many arrangements of clusters coexist as possible outcomes, which one the lattice reaches depends on where it starts, and it can wander for a long time before it settles. Glassy is named after this phase, not after its timbre; it is also where the device sounds most varied.

Phase map

The boundaries between the phases are not straight lines. They bend with a, and narrow windows of order appear inside the turbulent region. Glassy sounds most varied near these edges: try a = 1.8 with epsilon between 0.12 and 0.20. The map below counts the clusters left after 5000 steps for each pair of values.

Neighbours on a ring

The second model is Kaneko's coupled map lattice, from 1984: each element feels only its neighbours on a ring. Here the phases are spatial patterns: frozen random domains, selected stripes, spatiotemporal intermittency, travelling waves, defects that wander like Brownian particles, and fully developed spatiotemporal chaos. The manual lists settings for each.

Drift, spread and input

Drift lets epsilon wander by itself, so the lattice crosses phase boundaries without you touching it. Spread gives each cell its own a, which keeps the lattice from settling. An optional input can pull a block of cells toward the signal or move their a. The space-time view on the left of the device shows what you hear: stripes for cells that hold their values, bands for the whole lattice moving together, speckle for turbulence. Rec records the output straight into the Arrangement.

To sum it up

• Two models from Kunihiko Kaneko: the globally coupled map (GCM) and the coupled map lattice (CML)
• Up to 4096 cells, stepped at audio rate with its own clock (44.1k–96k)
• Drift, spread and loop for slow change or repeatable phrases
• Three readouts: one cell per side, the average, or the spread of each half
• Optional input that pushes the lattice or modulates a
• Space-time view of the lattice
• Rec writes a WAV and places it in the Arrangement

Compatibility

• macOS — tested on Live 12.4 / Max 9.1.4
• Windows — not tested yet
• Ableton Live 12 Suite, or Live with Max for Live and Max 9

Glassy produces full-scale noise. Start with the volume down. The models are implemented from Kaneko's published equations; see the manual for the references.

CycleScope shows the waveform of an oscillator. It detects the pitch of the incoming sound and folds the latest cycles onto one, so the waveform stands still on the screen. The audio passes through unchanged.

Wave

In the wave view, the cycle starts where the fundamental rises through zero, so a sine starts at zero going up. Faint lines behind the averaged cycle are single cycles: when they spread, the oscillator varies from cycle to cycle. Ref draws an ideal sine, triangle, saw or square behind it, scaled to the measured fundamental.

Spectrum

The spectrum view puts the harmonics as dots on a logarithmic frequency axis. Anything not locked to the pitch shows up between them: aliasing, noise, detuned oscillators, sidebands from FM or sync.

To sum it up

• Wave view: the latest cycles folded onto one, locked to the detected pitch
• Spectrum view with harmonic dots, 20 Hz to 20 kHz
• Readouts: frequency, note, level, pulse width, even/odd balance, noise
• Ideal sine / triangle / saw / square to compare with
• Double-click for a large, resizable window
• Audio passes through unchanged

Compatibility

• macOS — tested on Live 12 / Max 9.1
• Windows — not tested yet
• Ableton Live 12 Suite, or Live with Max for Live and Max 9

Browse and recall Roland JV-1080 patches directly from Ableton Live. Pick a category, pick a patch, and the hardware switches. The patch you choose is saved with your Live set and sent back to the JV-1080 when you open the set again.

Features

• All 640 factory and user patches (PR-A/B/C/D + USER), tagged into 17 categories
• No track routing needed: pick your MIDI interface once in the Port menu (ports named "JV" connect automatically)
• Bank select and program change with the correct JV-1080 bank map
• Fetch reads which patch is currently on the hardware
• Set recall restores the patch when you reopen the Live set

Compatibility

• Ableton Live 11.1 or later with Max for Live
• macOS (tested on Apple Silicon)
• Windows / Intel Mac: untested (set Port to "Track (Live routing)")

This device selects patches; it does not carry notes or audio. To play and hear the JV-1080, use Live's External Instrument (MIDI To → your JV-1080 port, Audio From → its inputs) or monitor through your mixer.

Not affiliated with Roland Corporation. JV-1080 is a trademark of Roland Corporation.

Browse and recall Yamaha FS1R performances directly from Ableton Live. Search by category, click, and the hardware switches. The performance and five knobs are saved with your Live set and sent back to the FS1R when you open the set again.

Features

• The factory performance banks, searchable by category
• Five performance knobs: Attack, Release, Formant, FM (hardware CCs) and REV (reverb return via SysEx)
• No track routing needed: pick your MIDI interface once in the Port menu (ports named "FS1R" connect automatically)
• Fetch reads the unit's performance channel and settings over SysEx
• Set recall restores the performance and knob values when you reopen the Live set

Compatibility

• Ableton Live 11.1 or later with Max for Live
• macOS (tested on Apple Silicon)
• Windows / Intel Mac: untested (set Port to "Track (Live routing)")

This device selects performances; it does not carry notes or audio. To play and hear the FS1R, use Live's External Instrument or monitor through your mixer.

Not affiliated with Yamaha Corporation. FS1R is a trademark of Yamaha Corporation.