Voices and machines
The Sidecar renders up to 8 voices, one per Rytm voice: all 12 tracks can choose a Sidecar machine, and like the Rytm’s own pairs, RS/CP, MT/HT, CH/OH and CY/CB share one voice, so eight sound at once. A track on a Sidecar machine sends its trigs and knob values to the Sidecar, which plays the voice and returns it on the track’s own channel, through the track’s analog filter and amp.
All knobs are 0..127 with the Rytm’s fine resolution. List knobs (ENG, PAT, MOD) split their range into equal zones, one per item.
PLAITS
Section titled “PLAITS”The Mutable Instruments Plaits macro oscillator, curated to 17 models.
| Knob | |
|---|---|
| ENG | the model: VA VCF, PHASE DIST, WAVE TERRAIN, STRING MACH, CHIPTUNE, VA, WAVESHAPE, 2-OP FM, GRANULAR, HARMONIC, WAVETABLE, CHORDS, SPEECH, SWARM, PARTICLE, STRING, MODAL |
| HRM, TIM, MOR | Plaits’ HARMONICS, TIMBRE and MORPH |
| DEC, LPG | the decay and colour of the low-pass gate, pinged by each trig |
| TUN | semitones from the centre (64) |
| LEV | output level |
| Page 2 | TMD MMD FM AUX / FIN VEL GLD VOL (as the Rytm shows them): timbre and morph modulation, frequency modulation, the aux output, fine tune, velocity, glide and volume |
The drum models are left out (DIGIDRUM plays drums), and the six-op FM models are their own machine (6-OP FM).
A Juno-106 style poly-synth voice, our own implementation following the open AMY model of the 106.
| Knob | |
|---|---|
| PAT | the patch: the 106’s factory patches (A11..B88), or the project’s JUNO list when it has one |
| CUT, RES, ENV | the patch’s filter cutoff, resonance and envelope amount |
| ATK, DEC, REL | the envelope |
| LEN | the gate; 127 holds until the note-off or the next trig |
Choosing a patch writes its own slider values onto CUT..REL, so the knobs show the patch.
Presets (JUNO and 6-OP FM). The release’s starter content (sidecar-content-<version>.zip, already on the Pi’s SD card package) puts the Juno-106’s 128 factory patches in /JUNO/Factory/ and three DX7 banks in /DX7/ on the card, ready for LOAD TO PROJ. PAT steps through the project’s list, up to 127 patches, and can be p-locked. Add patches with LOAD TO PROJ from the SIDECAR folder in the Rytm’s sample browser (the /JUNO and /DX7 folders of .syx files on the Sidecar’s card), or with Save current on the web page’s Pool & presets tab. Without a project list, PAT shows the factory list (JUNO A11..B88, the FM ROM voices). The lists and the patch libraries live on each Sidecar’s own card: copy them when you move a project between a Mini and a Pi. hardware (PAT presets on the Pi)
Put Juno-106 patch dumps (.syx, single patches or banks) in /JUNO on the Sidecar’s card, browse them in the SIDECAR folder and LOAD TO PROJ to build the project’s JUNO list.
6-OP FM
Section titled “6-OP FM”Plaits’ six-operator FM engine playing DX7 voices.
| Knob | |
|---|---|
| PAT | the voice: from the project’s FM list, or else 96 built-in ROM voices |
| TIM, MOR | timbre and morph offsets on the voice as stored (64 = as stored) |
| DEC, LPG | decay and colour of a low-pass gate after the voice; DEC 127 turns it off so held notes sustain |
| TUN | semitones from 64 |
| LEV | output level |
| LEN | the gate, as JUNO’s |
Put packed 32-voice DX7 banks (4104-byte .syx) in /DX7. In the SIDECAR folder each bank is a folder of its 32 voices; LOAD TO PROJ on a voice adds it to the project’s FM list. Banks are found by their content, so renaming or moving a bank keeps the list working.
DAISYSP
Section titled “DAISYSP”Physical-model resonators from DaisySP, excited by the Rytm.
| Knob | |
|---|---|
| ENG | MODAL, STRING or RESON |
| STR, BRI, DMP | structure, brightness, damping |
| INP | the exciter: INT (internal), NAT (the track’s own analog voice), NOISE, or another analog voice’s output |
| EXC | exciter amount |
| TUN, LEV | tuning, level |
| Page 2 | EXCITER: OUT FB XLV XTN / XDC XP4 XP5 WID (as the Rytm shows them). OUT at 64 or above sends the resonance to MAIN; below, back to the voice |
With INP on an analog voice, an analog drum rings a Sidecar resonator.
DIGIDRUM
Section titled “DIGIDRUM”Machinedrum-style digital drum engines from the open digidrum project.
| Knob | |
|---|---|
| ENG | EFM-BD, EFM-SD, EFM-XT, EFM-CP, EFM-RS, EFM-CB, EFM-HH, EFM-CY, TRX-RS, TRX-CB, TRX-CL |
| B..H | the engine’s controls; their labels change with ENG (PTCH, DEC, …) |
The track’s sample layer stays and plays under the drum. The engine’s PTCH sets the pitch; the note is not used.
DSP SAMPLE
Section titled “DSP SAMPLE”A sample from the Sidecar’s card, played through a resonator.
| Knob | |
|---|---|
| ENG | MODAL, STRING or RESON |
| STR, BRI, DMP | structure, brightness, damping |
| MIX | dry sample against the resonator |
| FB | feedback |
| TUN, LEV | tuning, level |
| Page 2 (SD SAMPLE) | TUN FIN SMP MOD STA END LOP LEV |
- SMP picks the sample from the project’s Sidecar pool (0 OFF). Fill the pool with LOAD TO PROJ in the SIDECAR folder.
- MOD is the playback mode, a stepped list (one item a detent): ONE, LOOP, SLICE, GRAIN, STRCH, PLUCK, WAVE. The STA, END and LOP labels change with it (STA END SLC for SLICE, POS SIZ DNS for GRAIN, and so on).
- On the Pi, samples play from memory (256 MB by default). On a Teensy without PSRAM, samples stream from the card and play once (ONE), or loop with LOP above 0; MOD is ignored.
DSP SAMPLE’s GRAIN mode: a granular voice over the pool sample, from memory, with position, size and density on STA END LOP. Pi, or a Sidecar Mini with PSRAM; on a Mini without PSRAM DSP SAMPLE streams from the card and plays once (ONE), whatever MOD says. Each Pi render core can sound up to 32 grains. GRAIN as a machine of its own in the Rytm’s SRC list is not available yet (the list is full): use DSP SAMPLE’s MOD.
For a stand-alone granular machine with its own built-in sample, see the GRAIN user machine in the starter set (User machines).
Measured on the Pi 4, each voice costs about 0.2 % (DIGIDRUM) to 1.2 % (6-OP FM) of one core, so eight voices leave most of the Pi free. On the Teensy, eight PLAITS voices average about 5 % each, so the Teensy’s budget is the limit for heavy models (MODAL peaks near 10 %). A voice that runs over its budget, or crashes, is muted on its own; the others keep playing.