ODD-1 Reference

The ODD-1 works beneath notes. It scans an audio buffer and pulls short fragments — grains — from wherever its playheads sit and streams the grains into continuous sound. Load a sample, move playheads through it, shape the grains. Every parameter is on the panel, and every one can be modulated.

Microsonic synthesis, like granular, selects short regions of a buffer, typically milliseconds long. It shapes and positions each region as a grain, and combines the grains into one output stream. The term "microsonic" describes the level the engine works at.

Signal flows upward through three stages that map to the right side of the panel: samples (the buffers the engine reads), playheads (cursors moving through a sample, each with its own position, speed, and tuning), and grains (the micro-fragments generated at each playhead).

The instrument is built for experiment. Hold any knob and it opens as a modulation destination. Press a source and the connection is made. There are no fixed routings and no lists to scroll, so trying something takes a gesture instead of a menu dive.

Each of the six rows has a button with two labels, black above white. The LED beside the button picks which label is live. LED off is the first (black) function, and LED on is the second (white). Six rows by two functions makes twelve panels. Two knobs sit outside the rows: VOLUME (far left, global) and ADJUST, which edits whichever AUX or secondary parameter is selected on screen. Left column, top to bottom: REVERB/SYSTEM, ECHOES/LAYERS, FILTERS/SATURATION. Right column, top to bottom: GRAINS/MACROS, PLAYHEADS/LFOS, SAMPLES/ENVELOPES. The right column's second functions are the modulation sources.

On the lower rows the knobs are numbered. Press one to select that instance, and the LED confirms the selection. All row controls then apply to that instance. The right side has four instances each of SAMPLE, PLAYHEADS, ENVELOPES, and LFOS. The left side has two each of FILTERS, SATURATION, and ECHOES. LAYERS is a fixed two-layer architecture.

Touching a parameter's on-screen knob momentarily overlays its patched modulation sources and an UNPATCH control. Press and hold instead to enter modulation patching mode, making that parameter a destination.

The upper half is a live visualization of the active samples and grains. It shows a spectrogram by default. Long-press it for a waveform. The four playheads each appear in a distinct colour (blue, green, pink, teal). The lower half shows the on-screen panel for the last-touched row. Every panel has the same layout: a small plot, a switch above it, buttons to its left, and on-screen versions of the row's knobs along the bottom.

Press AUX on screen to reveal a row's AUX parameters (press again to hide). The set is context-dependent and changes per active row. To change one, touch it. The parameter and the ADJUST knob turn red. Then turn ADJUST. Exit by pressing ADJUST, re-tapping the control, or switching rows.

Every knob is an endless encoder with detents around its whole rotation, so no knob has a centre you can feel. Four values hold on the screen for longer than usual as you turn through them: PLACEMENT at centre, TUNING at unison, POSITION at the start of the region, and SPEED at unity. The firmware calls these soft detents.

VOLUME in the upper left is the instrument's output level and has no second function. Pressing it opens the audio settings: input source and gain, and headphone level.

ADJUST sits beside VOLUME in the REVERB/SYSTEM row, and its two printed jobs follow that button: AMOUNT (reverb amount) with the LED off, PATCH (the active patch) with it on. It also sets every deeper on-screen value. Whenever ADJUST and a screen element both turn red, turn ADJUST to set the value and press it to confirm.

A knob's on-screen popup carries a RESET button. The button returns that one parameter to its default value. The layer-wide RESET on the LAYERS AUX panel returns every parameter at once.

A 6–15 V centre-positive 5.5 × 2.1 mm barrel jack (a 12 V DC adapter is included) feeds DC IN. Press ON/OFF to boot. Booting takes about 20 seconds, with an LED sequence and the ODD logo. Hold ON/OFF to power down.

USB-C presents a single device, "ODD-1", carrying both audio and MIDI. The same port mounts the unit as a USB drive for samples, patches, and firmware via SYSTEM > MISC > USB DISK.

Line-level stereo in and out sit on separate 3.5 mm TRS jacks. Audio is sent to all outputs at once. The headphone jack is 3.5 mm TRRS, supporting headphones and mic-enabled headsets. Input source and gain are found in SYSTEM > AUD. The same page selects how the input is read: stereo, left, right, or BAL for a differential or balanced mono source. Every connection is on the rear, left to right as the top edge of the panel labels it: headphones, AUDIO OUT, AUDIO IN, the USB-C port, DC in, the power switch, MIDI OUT, MIDI IN, and CONTROL SIGNALS A to D. Nothing sits on the sides.

MIDI arrives over USB-C and over 3.5 mm TRS MIDI IN (both TRS-A and TRS-B). The unit answers both at once, so use one type per source. Settings are in SYSTEM > AUD. MIDI also leaves the unit. TRS MIDI OUT can be set to MIDI-A or MIDI-B. MIDI Through passes input from USB and TRS out to the TRS jack. The MIDI jacks are on the rear, between the power switch and the control-signal inputs.

Four rear jacks (3.5 mm, 0–5 V) are labelled A–D left to right on the top right of the unit. They take modular gear within that range, or expression and sustain pedals through an adapter. Each is normalized to ±1 and available as a modulation source anywhere on the ODD-1.

The right side of the panel generates sound (Sample → Playheads → Grains). The grain stream crosses to the left side, which processes it (Saturation 1 → Filters → Saturation 2 → Echoes) and sends it to a shared Reverb. A layer is one complete instance of this chain. The ODD-1 has two identical layers. The LAYER ORDER button sets whether the layers run in parallel or in series. In parallel, each layer is fed from the input selector and the outputs are summed. In series, Layer 1's output feeds Layer 2's input.

The badge on an entry says how far its setting reaches. Grain is one fragment of sound, with its own filter and its own place in the stereo field. Playhead is one of the four cursors that move through a sample. Its level, start point, speed and tuning are its own. Voice is one played note. Each voice is saturated and filtered on its own before the layer sums them. The ODD-1 plays eight voices at once. A layer is a complete instrument, and the ODD-1 has two of them. With both playing, each layer has four voices. Global is what no layer owns exclusively: reverb, master volume, input level, and the control-signal inputs.

Five steps, each assuming the one before. A MIDI controller helps but is not required. The REVERB/SYSTEM button enters and exits SYSTEM. Most panel buttons toggle two functions, and the LED tells you which is live.

Press REVERB/SYSTEM and touch BANK to open the patch bank, which holds up to 16 patches. Scroll through them with ADJUST (labelled PATCH for this job) and press it to switch. To begin fresh, turn ADJUST to an empty position and press it. Press REVERB/SYSTEM again to leave SYSTEM.

Press ECHOES/LAYERS (press again if the ECHOES panel shows instead) to reach LAYERS. Press knob 2 and make sure its activity (speaker-icon) button is off to mute Layer 2. Then press knob 1 to work in Layer 1 for this walkthrough.

Press SAMPLES/ENVELOPES with the LED off and knob 1 to select Sample 1. Touch the sample name to open the browser. ODD is factory content and USER is yours. Navigate bracketed folders with ADJUST, and press ADJUST to load.

With a sample loaded, tap the sound visualization and work across the rows. Reshape the region with POS, LP ST, LP LN, and LEN. On PLAYHEADS, turn SPD down to slow Playhead 1, then raise Playhead 2's LVL and set its TUNE to −12 for an octave below. On GRAINS, explore SIZE, SPC, PLC, SHP. On the left side, try FILTERS' CUT and RES and ECHOES' AMNT and TIME.

Press REVERB/SYSTEM, touch BANK, select FILE then SAVE, name it on the on-screen keyboard, and press the check mark. The patch goes to the USER library and can be loaded into any bank position. Press REVERB/SYSTEM to exit.

The SAMPLES row (i.e., the bottom row on the right, LED off) controls four sample buffers per layer. Press knobs 1–4 to select which buffer to work with. The LED confirms the selection. All SAMPLE-row controls then apply to that buffer. By default a sample plays at natural pitch and loops continuously on middle C. It is re-pitched across the keyboard while timing stays constant. Harmonies can come easily from playheads at different tunings.

Touch the sample name in the header to open the Sample Browser. ODD holds the factory content and USER holds your own. The browser reads folders of .wav, .mp3 and .flac. Move up and down through the folders with ADJUST, auditioning as you go, and press ADJUST to load. Cancel returns you to the sample you were using before you started browsing.

Each layer holds four stereo sample buffers of up to sixty seconds each. There are eight buffers across the instrument. The ceiling is per buffer, not shared. A short one-shot in Sample 1 buys nothing for Sample 2.

In SYNTH the four buffers are not independent parts. Any playhead can be pointed at any sample (touch the assignment in the PLAYHEADS header and turn ADJUST). The playheads pull grains from wherever they sit, and everything sums before the amplitude envelope and the rest of the voice. The layer is one instrument fed by up to four sources.

Press a sample's knob and touch the RECORD icon to arm the buffer: ADJUST flashes red and an input meter shows. Press ADJUST to start and again to stop. An input threshold can start the recording instead. Trim what you caught with POS and LEN. Name it on the on-screen keyboard and it is placed in the USER library, available to any other patch. Recording does not depend on input monitoring. A mono source read as stereo leaves one side of the take empty; set IN CH in SYSTEM > AUD to match the source.

Samples added over USB go in the samples folder and may be .wav, .mp3 or .flac. Eject the drive from the computer before unplugging. Pulling the cable while mounted can corrupt the library.

POS adjusts the region of the sample used to generate sound, shifting the whole active window — the LEN region and the loop together — forward or back through the buffer. Every playhead assigned to that sample reads from this region, so POS offsets them all equally.

LEN sets the length of the active sound-generation region relative to POS. LEN is a hard limit for the loop: LP ST and LP LN operate within it. If LP LN would run past LEN's right edge, the loop stops at that edge. If LP LN is shorter, it governs loop length as expected.

LP ST sets the start point of the looping region inside the active window. In ENV LP a playhead that enters the looping region stays there until the note is released. In LOOP a playhead stays there regardless of note-off. Setting LP ST some way into the sample creates a region a playhead crosses on its way to the loop. This pre-loop zone can catch a transient before the loop begins.

LP LN sets the length of the looping region, working within the LEN ceiling. Length quantize keeps LP LN aligned to the layer tempo. Where LP LN would exceed LEN, it is clamped to the space that remains.

PITCH sets the base pitch used to calculate the shift for each incoming MIDI note. The ODD-1 analyses the audio for a dominant pitch and sets PITCH to it automatically. With none or several detected, PITCH defaults to C4. The sample then plays naturally on middle C. Turn fully left for NONE to disable note-based pitch shifting entirely.

TEMPO sets the sample's native tempo for sync purposes. The value is taken automatically from the filename or audio analysis. TEMPO is NONE when no tempo is detected. Speed BPM-ratio sync locks playback speed to TEMPO. Press TEMPO to reveal TAP TEMPO.

MODE selects how the sample plays back. 1-SHOT stops playheads at the sample boundary without auto-stopping the note. ENV LP keeps playheads in the LP ST / LP LN region while a note sustains and releases them on note-off. LOOP holds them in that region unconditionally, ignoring note-off.

Press and drag on the waveform to zoom (up and down) and navigate (left and right). A long press on the upper display switches between waveform and spectral overviews. Turning LEN reveals a CROP button that trims the sample to the currently selected region.

Loading a sample runs pitch analysis. A clear single pitch sets PITCH to the nearest MIDI note, so samples recorded at different pitches all play in tune from the same key. When no single pitch is found, PITCH is set to C4 and the sample sounds naturally at middle C.

Loading a sample into a free buffer assigns the first silent playhead to it and raises that playhead's level. A second sample therefore sounds immediately alongside the first, without any routing step.

The PLAYHEADS row (i.e., the middle row on the right, LED off) gives up to four playheads per layer, eight in total. Each playhead marks a position in its assigned sample from which grains are continuously generated. Each can be assigned to any of the four buffers with its own position, speed, tuning, and level. Select one by pressing its numbered knob. A playhead can also be assigned NONE, leaving it unused rather than silent.

LVL is the output level of grains generated by this playhead — its contribution to the layer output. All four playheads are then shaped together by the one layer amplitude envelope, on ENVELOPES knob 1, and the effects chain downstream.

STRT is the playhead's position in the sample when a note is triggered. The playhead resets to STRT on each new note. Retriggering re-syncs every playhead to STRT. A new note can begin while the previous one is still fading.

TUNE sets the relative pitch of grains generated by this playhead — the interval it plays relative to the incoming note. Setting different playheads to different TUNE values stacks intervals into harmony from a single sample.

SPD is the rate at which the playhead travels through the sample, from frozen stasis at zero to near-conventional playback when high. SPD is independent of pitch. A playhead can crawl through the material while still sounding at its TUNE pitch.

Every playhead and every sample carries a SOLO button. Soloing narrows what you hear to that one source without touching any level. A playhead can be picked out of a dense texture and put back without losing the balance you built.

MID CH assigns a MIDI channel to this playhead, so playheads can be driven from different channels or note ranges independently rather than all following the layer channel.

L-R MODE sets what this playhead does with the sample's two channels. SRC keeps the sample's own format, mono or stereo, as it is. MONO sums a stereo sample to mono before the grains are placed in the stereo field. Use it when the binaural effect matters more than the sample's natural image. L-R MODE also lists LEFT and RIGHT.

With REL MD on, a playhead's TUNE and SPD become offsets from Playhead 1's and follow it. Set PH 2 TUNE to +12 and it stays an octave above wherever PH 1 goes. Relative mode covers every playhead parameter. LVL, STRT, TUNE and SPD all follow Playhead 1 with an offset, or stay fully independent. The playhead modes follow too. Playhead 1 is always the absolute reference the others can track.

Beyond the layer's own MIDI settings, each playhead can take its own MIDI channel and key number, where the layer's CHANNEL mode and KEY RANGE allow it. A layer's key range bounds its playheads' ranges. A drum pad or sequencer can then trigger playheads one at a time. A single layer then becomes four separately addressable voices reading the same buffers.

JUMP lives on the pop-up behind playhead START. JUMP scatters the read position for each grain, so repeated material stops sounding like a loop. It leaves grain timing alone. SPACING JITTER on the GRAINS row randomizes grain timing.

QUANTIZE TUNING constrains tuning to the layer's SCALE — both the values you dial and the modulation arriving at them. A detuning LFO or an envelope sweeping pitch then lands on scale degrees rather than gliding between them.

A playhead reaching the boundary of its region follows a boundary mode. In ZERO mode the playhead falls silent at the boundary. It sounds again if it re-enters the region. Grains already in flight play out rather than being cut off.

The GRAINS row (i.e., the top row on the right, LED off) shapes the fragments each playhead generates. Every primary grain parameter has a paired secondary. Most secondaries add randomization, giving each playhead more than 20 grain controls. Press a primary's physical knob to reveal its secondary as centre and spread values. The header shows CONTROL MODE: touch to toggle ALL PLAYHD (every playhead at once) or SEL PLAYHD (only the selected one). In SEL PLAYHD, touching a playhead's grain on the lower subplot makes that playhead the active one.

SIZE sets the length of each grain in time. Short grains read as crunchy and noise-like, and long grains read as smooth and timbral. With SPD and SPC, SIZE forms the core texture and gives timbral weight for whatever motion SPD establishes.

SPC sets the interval between successive grain onsets. Wide spacing gives separated, rhythmic pulses, while narrow spacing gives a dense, overlapping cloud. SPC is the density control, best set last, after SPD and SIZE.

PLC sets the perceived position of grains in the stereo field, using a proprietary perceptual panning algorithm. With its PLACEMENT SPREAD secondary it can scatter grains across the stereo image, widening a single playhead into a spatial texture.

SHP sets the window applied to each grain, from a bell curve to a rectangular window. Softer windows blend grains smoothly, while harder windows give a more percussive, edged grain. Press SHP to reach its secondary function, SHAPE BIAS, which tilts the window toward its front or back.

SPC is not an absolute interval. SPC is measured against the grain's length after re-pitching. At 0% a grain begins every cycle, and the grains fuse into that pitch rather than reading as separate events. Near 50% grains overlap smoothly. Toward 100% the stream turns sparse. This is why the same SPC setting sounds different an octave up.

A grain runs from a millisecond or two up to five seconds. SIZE is shortest at the knob's centre: turn right for long forward grains, left for long reversed ones. Below roughly fifty milliseconds grains stop being audible events and start being tone.

SIZE set against SPC fixes how much grains overlap, with three audibly different results: a smooth continuous stream, a pulse you can count, or pitched fusion when the grains lock to the played note's cycle.

Touch the on-screen SPC knob and turn SYNC on, and the interval between grains locks to the layer's tempo rather than to the played note. Grains then stay in step across different pitches, so rhythmic grain patterns hold together under a melody.

The four primary controls are one calculation, run per grain. SIZE fixes the length taken from the source. TUNING divides it to give the length after re-pitching, so pitching up shortens the output grain and pitching down lengthens it. SPACING multiplies that output length to set the gap to the next output grain. SPEED multiplies that gap in turn to set how far the read position advances in the source. On the panel, SIZE and SPACING sit on the GRAINS row and TUNING and SPEED on the PLAYHEADS row. The screen abbreviates them SPC, TUNE and SPD, and entries that describe the screen use those forms.

The paired secondaries randomize the primaries around their set value. SIZE VARIABILITY spreads grain size. SPACING JITTER spreads the interval between grains. A little of each loosens a mechanical grain stream into something organic. Reveal them by pressing the primary's physical knob.

PLACEMENT SPREAD randomizes grain position across the stereo field. SHAPE BIAS biases the window shape toward front or back. Together with the panning of PLC they turn one playhead into a wide, moving field rather than a point source.

The GRAINS AUX parameters randomize individual grains. DROP randomizes their level, JUMP their position, and DTUNE their pitch. RPT randomly repeats the previous grain. Each is an independent, stackable layer of variation. RPT in particular adds density without raising the grain rate.

Most grain parameters have a randomizer — a secondary or an AUX parameter — that varies them per grain. Used together they turn a single setting into a region. Instead of every grain sharing one size, spacing and position, each grain is drawn from a spread around them. The texture does not repeat grain for grain.

Grains over 100 ms, spacing low with some JITTER, and some JUMP on the playhead's START pop-up. Then set the secondary playheads to relative mode: press AUX and switch REL MD on. Bring their levels up and detune them slightly against Playhead 1 for thickness.

The FILTER panel inside GRAINS reveals an optional per-grain band-pass that filters each grain individually before grains are mixed. The ODD-1's other filter, the voice-level FILTERS row, processes the summed output downstream. The two can be easily confused.

BW VAR randomizes the filter's bandwidth per grain. C VAR randomizes its centre frequency per grain. Each grain is therefore filtered a little differently. C VAR alone produces a timbral shimmer. Combined with BW VAR, the effect becomes fuller and less predictable.

The FILTERS row (left, LED off) applies two resonant filters per voice. Press knob 1 or 2 to select one. The header sets the type: low pass, high pass, band pass, band stop, or comb. The panel plots each filter's response and the combined curve in real time. This is the voice-level filter, distinct from the per-grain band-pass inside GRAINS.

CUT sets the filter cutoff frequency and RES its resonance. GAIN (AUX) sets the output gain of the filter stage. All three apply independently per voice, so each note is shaped before the voices are summed.

The SATURATION row (left, LED on) gives two saturation units. Saturation 1 operates per-voice, before the filters. Each voice is driven individually. MIX blends processed against dry. DRIVE sets the severity. IN/OUT (AUX) set input and output gain. The type is DISTORT, BIT CRUSH, or COMPRESS.

Saturation 2 processes the summed layer output after the filters. The whole layer is driven as one, rather than voice by voice. It shares Saturation 1's controls (MIX, DRIVE, IN/OUT, and type). Only its position in the chain differs. The same module appears at two scopes.

The ECHOES row (left, LED off) gives two echo units per layer, placed after Saturation 2. Press knob 1 or 2 to select one. Each runs in STEREO or PING PONG mode, set from the header with ADJUST. The pair can be wired in series or parallel with the 1|2 button. AMNT sets the echo output level and TIME the interval between echoes.

The echo AUX parameters mostly act inside the delay loop. FDBK sets the feedback amount, DRV the drive applied within the loop, FILT a band-pass filter on each repeat, and IN GN the gain going into the unit. Together they take echoes from clean digital through warm analog, and on into chorus, flanger, or runaway feedback. BYPASS sits in the visualization panel.

The REVERB row (left, LED off) is a single reverb shared by both layers at the end of the signal chain. The reverb is global, not per-layer. Each layer feeds the reverb independently via its SEND control in the LAYERS row, so the two layers can sit at different depths in the same space. AMNT sets the overall reverberation level. A layer's LEVEL and SEND combine so that any wet/dry balance is reachable, including a layer heard through the reverb alone.

The reverb AUX parameters shape the tail. TIME sets the decay length. SIZE sets the size of the reverberant space. P DLY sets the pre-delay before the reverb arrives. CLR sets the colour, from dark to bright.

The ENVELOPES row (right, LED on) gives four envelopes per layer: three standard ADSR envelopes plus TIME PATH on knob 4. Envelope 1 (AMP ENVELOPE) drives each note's amplitude by default but can also modulate any destination. Envelopes 2 and 3 have no default connection and are free as modulation sources. All three can run at once and stack on one destination. Their contributions add up to the destination's ceiling.

The four knobs set attack time, decay time, sustain level, and release time — the shape of the envelope over a note. The header toggles the segment shape between LINEAR (straight segments, precise for modulation ramps) and LOG (curved segments, suited to amplitude).

SCALE, FLIP and LOOP are common to all three envelopes. SCALE scales the envelope's maximum output from 1 down to 0. FLIP inverts its output. LOOP cycles it through all ADSR phases. On Envelopes 2 and 3 the sustain length used while looping is adjustable.

Envelope 1's AMP setting enables or disables its connection to note amplitude. Turn it off and Envelope 1 is freed as a pure modulation source while its release time stays active. Turning AMP off is the way to keep a sample's own dynamics. Envelope 1's velocity sensitivity can also be switched off in the same AUX panel. Envelope 1's ENV1A button opens the amplitude connection's destination list. Envelopes 2 and 3 show MOD DESTN instead, with UNPATCH to remove their connections.

TIME PATH (knob 4 of the ENVELOPES row) is a dedicated envelope that controls playhead speed dynamically over a note rather than amplitude. Its plot looks like an ADSR, but the vertical axis is sample position. The slope is therefore speed: rising is forward, flat is frozen, falling is reverse. Every note starts fresh from STRT and follows the same journey.

The four knobs set the relative playhead speed for the attack, decay, sustain, and release phases. For instance, a time path can sweep quickly through a sample at onset and settle to near-stillness at sustain. These per-phase speeds are relative to the PLAYHEADS SPD control.

A TM and D TM set the duration of the attack and decay segments of the time-path envelope. RESET returns all time-path parameters to their defaults. The visualization plots a coloured time path per active playhead. Flat segments mark where a playhead is frozen in the sample.

Set Envelope 1's release long. Ten seconds is not too much. Press knob 4 for TIME PATH and take its release speed to near zero. Let a note go and the playhead stops where it was, ringing out from that single instant for the whole release.

The LFOS row (right, LED on) gives four LFOs per layer. None is connected by default. The header shows the waveform. Touch it and use ADJUST to pick SINE, TRIANGLE, SAWTOOTH, REVERSE SAW, SAMPLE & HOLD, or SQR. MOD DESTN opens the destination list, with an UNPATCH that removes all of the LFO's connections.

RATE sets the LFO frequency, from a slow drift to an audible flutter. DPTH sets the maximum amplitude of the LFO's output. That amplitude is how far the LFO pushes a destination.

BIAS tilts the output toward +1 or −1. WARP modifies the wave shape for most types, sets the duty cycle for square, and sets the sequence length for sample-and-hold.

MODE selects free-running or note-triggered. In note-triggered mode a time-delay parameter lets an LFO start late in the note. PHASE sets the starting phase of the waveform. QUANT quantizes the output into steps. SMTH smooths the output.

Sample-and-hold is one of the LFO types. Instead of a continuous shape it holds a new random value at each cycle, giving stepped rather than swept motion. On SAH alone, WARP changes the number of steps. Lower WARP for fewer, larger jumps.

Tap an LFO's on-screen RATE knob and SYNC appears. Turned on, SYNC locks the LFO's rate to the layer's tempo. RATE becomes a musical division. A stepped filter sequence or a tremolo then stays in time instead of drifting against the tempo.

The MACROS row (right, LED on) gives four manually adjustable sources: A, B, C, and D. Each connects to any number of destinations at once. Unlike LFOs and envelopes, macros move only when you turn them. A single macro turn can move several parameters in a designed relationship. The macro panel has its own layout of four columns, one per macro. Pressing a macro's encoder opens that macro's own panel.

On first connecting a macro, the direction you turn it sets polarity: clockwise positive, counter-clockwise inverted. One knob can therefore open a filter while shortening grain size. MOD DESTN lists a macro's destinations, and UNPATCH disconnects them all. Long-press an on-screen A, B, C, or D to see everything that macro currently drives.

A layer has up to 24 modulation sources. Eleven are on the front panel: 3 envelopes, 4 LFOs, and 4 macros. The rest are 4 control-signal inputs, 7 MIDI sources, and 2 touch-screen dimensions. Almost any can be connected to almost any parameter.

A connection links a source — envelope, LFO, macro, MIDI, or control signal — to a parameter. You start by opening the parameter as a destination — hold its knob — then reach for the source. Every connection has a strength (a signed percentage) and a direction. Connections are per-layer and save with the patch.

Press and hold a parameter's on-screen knob and the display switches to the MOD SOURCES screen with that parameter as the destination. The parameter's name appears in the MODULATE panel at the bottom. For an AUX parameter, press the row's AUX button, tap the AUX parameter (it highlights red), then press and hold ADJUST. Not all AUX parameters can be destinations.

The MODULATE panel sits at the bottom of the MOD SOURCES screen. Its header names the open destination. The left side lists the sources connected to that destination, with an UNPATCH that removes them all. The right side shows the connection strength as a percentage, set with the adjust knob. EXIT leaves patching mode and returns to the previously active row.

On-panel sources are reached from the hardware rows while MOD SOURCES is open. They do not appear as on-screen points. Press a source's knob to connect at full strength. A grey arc on the destination shows the mod range. Turn the knob instead to connect at a variable strength that updates live. Turning left past zero gives a negative, inverted connection.

MIDI and control-signal sources are connected from the MOD SOURCES screen. Tap a point to connect or disconnect it. Active points highlight. Select a point and turn ADJUST to set its strength, shown in MODULATE. Turning past zero gives a negative connection. CTRLS A–D are the four rear control-signal jacks.

The X-Y pad at SYSTEM > XY is a modulation source like any other, offering two dimensions that can be patched to separate destinations. The pad is the pair of touch-screen sources counted in the layer's twenty-four. It is reached from the display rather than the panel rows.

To remove every connection to a parameter, touch its on-screen knob and press UNPATCH in the column, or hold its physical knob and press UNPATCH on the MOD SOURCES screen. To remove a single connection, press an on-panel source's knob while the destination is open, or tap an on-screen source's highlighted point.

Press AUX on the LAYERS page: its UNPATCH removes every modulation connection for the current layer at once, leaving the other layer's connections untouched. It clears only the routing and leaves each parameter at its current value. The sound holds but stops moving.

The four rear control-signal inputs (A–D) are available as modulation sources throughout the ODD-1. Each input is configured in SYSTEM > CTRL. Each is normalized to ±1 based on the voltage range you set. External CV, an expression pedal, or a sustain pedal can then drive any destination.

MIN and MAX set the expected voltage range, normalized to ±1. BIAS shifts the normalized output up or down. MODE selects normal, inverted, or switch. Switch mode is off by default. In switch mode the input acts as a toggle. Each time the processed signal crosses 0.5, the output flips between 0 and 1.

Four MIDI signals form the MIDI 1 group of sources: KN (key number), VL (velocity), MW (mod wheel), and CC (assignable control change). Connect them from the MOD SOURCES screen. For CC, tap the point and move a controller to capture it. The point reads CC LEARNING until the controller responds, then shows the controller's name.

The MIDI 2 group adds PB (pitch bend), AT (aftertouch), and BR (brightness). All three support MPE, and AT also supports Poly AT. With an expressive controller these turn per-note gesture into modulation. Control-signal inputs A–D also appear on the modulation-sources screen under CTRLS.

Two MIDI controls reach past modulation. A program change message switches the active patch. Fixed control changes 110–117 switch panel selections. The looper uses the fixed CCs 118 and 119 for record arm and transport. CC 119 also freezes LIVE recording.

The ODD-1 has two layers. Each is an independent instrument with its own samples, playheads, grains, envelopes, LFOs, macros, filters, saturation, and echoes. Only one is under panel control at a time. Press knob 1 or 2 in the LAYERS row to select it. Reverb and master volume are the shared stages.

LEVEL sets the layer's output level and SEND sets how much of the layer goes to the shared reverb. Per-layer send lets one layer sit dry and close while the other is spacious and distant.

TEMPO sets the layer's BPM and updates automatically from MIDI clock when CLOCK is set to SYNC. Each layer has independent TEMPO and CLOCK. SCALE quantizes pitch changes to a musical scale, or NONE for unquantized operation.

Every MIDI field on the LAYERS page follows SYSTEM > MIDI until you change it on the layer, and from then on the layer owns that field for this patch. Tap the field's GLOBAL chip to re-attach it to the system setting. SET FROM PATCH does the reverse, making the current arrangement the new default.

The ODD-1 plays eight voices at once. A layer playing alone can use up to eight while, with both layers on, each has four. Every playhead responds to incoming notes. A layer is a polyphonic instrument, so two layers split across a keyboard work as a performance setup.

Four things are instrument-wide: the reverb, master volume, input level, and the control-signal inputs. Everything else belongs to the layer, including its MIDI settings and sync switches, which follow SYSTEM > MIDI until you change them on the layer itself.

A layer runs in one of three modes, selected with the mode button on the LAYERS page. SYNTH, the default, is the multi-timbral synth and sampler, with up to four playheads across four samples. LOOPS turns the layer into a musical looper, where notes trigger loops rather than re-pitching them. LIVE granulates incoming audio through the full engine in near-real time, with about 5 ms of delay. It uses a single continuously written buffer and is fully keyboard-playable.

Press MIDI in the LAYERS row for per-layer settings: CHNL (ALL, MPE, a specific channel 1-16, or NONE), LO KEY and HI KEY (the note range the layer answers to), and TRNSP and CLOCK to enable transport and clock sync. Overlapping ranges on one channel layer the two sounds, while non-overlapping ranges split the keyboard.

In LOOPS, with the first sample buffer empty, the layer goes straight to recording or loading. Playback starts on its own when recording stops. Notes trigger loops instead of re-pitching them. Play a note to start its loop, and play the same note again to stop it. Note-offs are ignored, so nothing needs holding down. Each loop is captured into its own sample buffer rather than overdubbed into a shared one. When recording stops, the layer moves you on to the next sample, ready for the next loop.

A loop plays at the Pitch value stored with its sample, and the pitch detected on load is not applied. Together with automatic speed sync, this keeps several loops of different origins locked together.

The layers do not have to share a mode. One can loop while the other is played as a synth, or a LIVE layer can granulate over a SYNTH bed. The LAYER ORDER button, just below the LAYER 1 / LAYER 2 label, switches the pair between parallel and series. In series, Layer 1's output feeds Layer 2's input, so the second can capture, loop or granulate what the first plays.

In LIVE, Sample 1 becomes a buffer that is continuously overwritten from the audio input, and all four playheads read from it. Instead of granulating a stored sample you are working on the recent past of whatever is coming in. The buffer is the full sixty seconds a sample slot holds. LEN sets how much of the buffer you work with and starts at about four seconds.

A buffer being overwritten has no start to point at, so the time controls change meaning. PLAYHEAD START and LP ST are lags from the present. A value of 0 ms is the moment being written, and turning the knob left reaches further into the past. LP LN is how much of that window plays. In LIVE, SAMPLE POS is replaced by a shortcut to the input monitor control.

There is no arming in LIVE because capture is always running. The record button freezes the buffer instead. What is in the buffer stays, and you keep playing it like a stored sample. Press again and the live input fades back in.

A pedal can arm and run the looper. Connect it by MIDI CC learning or through the control-signal inputs. With record arm active, hold ADJUST to patch record start and stop. Fixed CCs also work. CC 118 arms, and CC 119 starts and stops. CC 64 is reserved for sustain.

Press REVERB/SYSTEM to open the global panel — everything that applies across both layers — organized into sub-panels from the header: BANK, AUD, CTRL, MISC, and XY. Press SYSTEM again to exit and return to normal operation.

BANK holds sixteen positions in local memory, each a complete snapshot of both layers' parameters, mod connections, and settings. The BANK is the working set, distinct from the permanent patch library. Turn ADJUST through the list and press to load. A * before a name means the position has diverged from the library or was never saved to it. An empty position shows a dash, and an empty position you have started editing shows *-.

The BANK is the sixteen slots you scroll with PATCH, which hold what you are performing with now. The LIBRARY is every patch ever saved. Loading a library patch copies it into a slot. The slot remembers where the patch came from, so the instrument can tell you when your edits have diverged from the original.

EXPORT bundles a patch together with every sample it needs into a single file and makes it visible on the ODD-1's USB drive. It is how a patch travels — to another unit, to a collaborator, or between the hardware and the ODD-1V plug-in. EXPORT is offered while saving a patch (BANK, FILE, SAVE).

The AUDIO AND MIDI sub-panel selects the input source for LIVE granulation and recording: USB IN (no gain), LINE IN (activates IN GN), or MIC IN (mic via the TRRS headphone jack, activates MIC GN). The sub-panel also sets HP OUT headphone level relative to line out. Stereo meters sit at the bottom. IN MON is the input monitor, the level at which the input is heard directly at the output. Recording does not depend on the monitor level.

CLOCK SRC in SYSTEM > MIDI picks where incoming MIDI clock is read from: TRS, USB, or AUTO to follow whichever is sending. Each layer keeps its own tempo and decides how it responds through its own CLOCK and TRANSPORT switches.

Connect the BPM knob to AUX ADJUST and a tap tempo button appears on screen. Tap it in time with a tempo you are hearing and the instrument follows.

Touch an on-screen knob and its pop-up offers SYNC and QUANT where the parameter supports them. With one selected, a small metronome icon appears on that knob and shows at a glance which controls are following the clock and which are free.

The CONTROL SIGNALS sub-panel calibrates and configures the four CV inputs (A–D) as modulation sources. A summary shows all four with meter and depth. Touch a tab for one signal's detail. PLOT SOURCE toggles the plot between raw input (5V) and normalized value (±1).

MISC holds device utilities: INFO (serial, hardware and firmware versions for support), VIZ (an LED BRIGHT slider), DBG DUMP (a developer diagnostic, not normal operation), and USB DISK, which mounts the ODD-1 as USB mass storage for samples, patches, and firmware.

Over USB-C the ODD-1 appears as a stereo audio interface and a MIDI device, with no drivers to install. Record the instrument into a DAW, send audio back out to it, and play it from any sequencer or MIDI track on the same cable.

While the ODD-1 is mounted as a drive the screen locks and only the volume knob responds. Eject from the computer to return the panel to normal. The lock is deliberate and keeps the panel out of the way while files are moving.

Download the release package, mount the unit with SYSTEM > MISC > USB DISK, and drag the firmware sub-folder across. Eject, then follow the prompts on screen to update and reboot. The package also carries release notes, which are worth reading before you start.