Playback engine 15 of 15
Pulsar
Play a recording as a train of short, windowed fragments called pulsarets. Slow repetition gives separate sampled pulses; faster repetition becomes a pitched buzz with the recording’s changing character.
Good for: vocal reeds and growls, metallic tones, pitched machinery, interrupted rhythms and unstable pulse textures. Every pulsaret reads the recording; empty or silent sources stay silent.
Pages and navigation
Select Pulsar with ALG on Engine. It has two unique pages and the shared Engine, Pads & Scale, Amp Envelope and Stack Path pages. Quad omits Stack Path. Click the main encoder on the Pulsar page to hold/resume source travel. Shift + pad latches a note; HOLD = MAN keeps it latched indefinitely.
Pulsar — main parameters
| Knob | Display | Range | What it controls |
|---|---|---|---|
| 1 | RATE | 0.1–2000 Hz | Pulse repetition rate at Root. Note pads transpose this clock by musical intervals, independently for each note. The default 261.63 Hz supports C-based keyboard playing at the default Root. Slow rates expose individual pulses. The actual transposed clock is bounded to 0.05–2000 Hz. The dial uses logarithmic spacing for fine low-rate changes. |
| 2 | FORM | 20–2000 Hz | Pulsaret duration as an inverse frequency: 500 Hz gives a 2 ms fragment, 100 Hz gives 10 ms. Raising FORM compresses the chosen source shape into less time. Waveform-view knob 7 is a shortcut to FORM. It shapes the spectrum independently of RATE, without promising a particular resonant frequency for arbitrary recordings. Overlap limits can shorten the requested duration; the footer reports LIMIT. |
| 3 | SIZE | 0.5–100 ms | Source fragment span: how much recorded audio is traversed inside one pulsaret. Increasing SIZE packs more source material into the same output duration. Position and Field bound the available source; SIZE is also limited by that region. Its dial uses logarithmic spacing. |
| 4 | WIN | HANN / BELL / TAPR / PERC / REV | Window shape. Hann and Bell round the pulse; Tapered has a flatter center; Percussive emphasizes an early peak; Reverse reverses that envelope, rather than the sample. Every choice has smooth, zero-valued edges. |
| 5 | SKEW | 0–100% | Window time skew moves activity toward the start or end. 50% leaves the selected window’s timing unchanged. It preserves silent edges and applies to new pulsarets. |
| 6 | OMIT | 0–100% | Pulse omission probability. 0% keeps every pulse permitted by Burst; 100% launches none. The clock, source travel and stack path continue. Existing pulses finish their windows. |
| 7 | BRST | CONT / pulse:rest | Burst mask: Continuous, 2:1, 3:1, 4:1, 1:1, 2:2, 3:3, 4:4 or 8:8. For example 2:1 permits two attempted emissions, then rests for one. Omit can additionally remove permitted emissions. At audio rates, these groups change the spectrum and perceived pitch. Changing the mask keeps the running clock position. |
| 8 | JITT | 0–100% | Timing jitter varies emission intervals by up to ±45%. Zero preserves a periodic pulse train. Higher values introduce instability and noisier textures, and reduce the safe overlap duration. It does not randomize stereo channels independently. |
Pulsar Mod — movement and variation
Up/Down from Pulsar opens PULSAR MOD. The top row contains two sine-modulation speed/depth pairs. The lower row varies source position, pulse duration, reading direction and keyboard response. Each note has its own modulation phases, starting at note onset. Speeds and depths can be changed while notes are held.
| Knob | Display | Range | What it controls |
|---|---|---|---|
| 1 | ERAT | 0.01–20 Hz | Emission modulation speed. Slow settings produce acceleration/deceleration; faster settings produce vibrato and more rapid pulse-clock movement. Default 0.13 Hz. The dial and MIDI encoding use logarithmic spacing. |
| 2 | EDEP | 0–24 semitones | Emission modulation depth above and below the note’s base pulse rate, measured in musical intervals. Zero disables its influence. The clock retains continuous phase through modulation and manual RATE edits. |
| 3 | FRAT | 0.01–20 Hz | Formant modulation speed. Default 1.7 Hz. It moves the pulsaret duration without retuning the pulse clock. Dial and MIDI encoding use logarithmic spacing. |
| 4 | FDEP | 0–24 semitones | Formant modulation depth around FORM. A rise of twelve semitones doubles the effective FORM frequency and halves pulsaret duration. Each new pulse captures the current modulation value; sounding pulses finish their windows. |
| 5 | SPRY | 0–100% | Source-position scatter. New pulses choose a displaced origin up to half the Field width in either direction at maximum. Selection stays inside Field. With source hold on, the held point remains the anchor while new fragments can scatter around it. Zero uses the exact current or held source position. |
| 6 | FVAR | 0–24 semitones | Formant variation chooses an additional random FORM offset for each pulse. Unlike JITT, it changes pulse duration and spectral character while leaving emission timing alone. It combines with FDEP and KEYT. |
| 7 | REV% | 0–100% | Reverse probability. Zero reads every fragment forward; 100% reads every new fragment backward. The chosen direction remains fixed for the pulse. Both stereo channels use the same direction. WIN’s REV setting separately controls the amplitude envelope. |
| 8 | KEYT | 0–100% | Formant keyboard tracking relative to Root. At zero, FORM stays fixed across played notes. At 100%, an octave rise doubles effective FORM as well as the note’s pulse rate. Intermediate values give partial tracking. It changes the timbral response while retaining the existing note-to-clock mapping. |
All new depths and amounts default to zero. The two modulation speeds run independently of Speed and stack scanning. Effective FORM stays within 20–2000 Hz and the pulse clock within 0.05–2000 Hz, so extreme settings can flatten modulation at those bounds. Existing overlap limits remain active.
Try it: for a moving vocal texture, hold a source and start with FRAT 0.2–1 Hz and FDEP 3–7 semitones. Add a little SPRY for changing articulation. For unstable pulses, try FVAR 2–5 semitones with REV% around 25%. Add EDEP gradually when you want the pulse rate or pitch to move.
Notes, sources and lanes
Keyboard notes transpose the pulse clock relative to Root. The sample’s own shape remains audible inside each pulse, and stronger harmonics can dominate the perceived pitch. Speed controls source-origin travel; it does not transpose the pulse clock. Position and Field choose the usable region. Stack Start, Scan, drawn paths and Loop’s stack hold keep their familiar roles.
Each new pulsaret captures the lane blend, source span and window settings. It finishes that window while later pulses can use the next lane or new settings. The waveform retains each note’s source cursor and stack marker; a short bracket marks its latest launched fragment on the active lane. A small activity mark pulses at slow rates and stays lit for active audio-rate trains. The display does not attempt to draw every audio-rate emission.
Hold a source
On Pulsar, click the main encoder to hold source travel. The footer reads SRC HELD CLICK: RUN. It captures the newest active Pulsar note’s position in that part, favoring held notes; with no active note it uses Position. New notes reuse the held point while keeping independent clocks. Cues can move it. Click again to resume travel. Stack scanning and pulse emission continue independently.
Source hold, its position and all sixteen parameters save with clips. Source-hold changes record as discrete events. Pulsar can be an entry in an Engine List. Single retains eight note voices, Split/Shared Dual four per part and Quad two per part.
Overlap and live edits
Each note has four fixed internal pulse readers. Requested duration is limited to about 2.9 of the shortest current emission intervals, with a sixteen-sample minimum. The footer shows LIMIT: RATE / WIDTH when this shortens a pulse or a live acceleration temporarily fills the reader pool. Increase FORM, reduce RATE, or reduce JITT to ease the limit. A full pool omits a new emission while existing pulses finish.
RATE, FORM, SIZE and modulation amounts glide over about 25 ms; each sounding pulsaret retains its launch settings. Large changes therefore settle through upcoming pulses. Stereo channels share their source selection, windows and filtering. Prepared source filtering enters gradually; a bounded fallback permits immediate playback while analysis is pending. Overlap gain control and DC removal condition the pulse train before the existing note envelope and effects. Short, interrupted pulses can sound deliberately rough; these safeguards address abrupt reader changes and uncontrolled overlap.
Starting points
Vocal reed: hold a sustained vowel, use RATE 261.63, FORM 500–1200, SIZE 2–8 ms, Hann, and zero Omit/Jitter. Play notes, then move FORM.
Sampled percussion: use RATE 2–12 Hz, FORM 20–100, SIZE 20–100 ms and Percussive. Try BRST 2:1 and a little Omit.
Unstable machinery: use RATE 80–300 Hz, a short SIZE, Tapered or Reverse, BRST 3:1, and raise Jitter gradually. Move through the lanes for changing material.
MIDI and E16
Pulsar is algorithm index 14. NRPN 3328–3335 follow the eight knobs on channels 5–8 for A–D. RATE, FORM and SIZE use logarithmic 14-bit value encoding to retain precision at the low end. Source Hold is 3336 (Off/On); held position is 3337 (0–1). Pulsar Mod uses 3338–3345 in knob order. ERAT and FRAT also use logarithmic 14-bit encoding. E16 ENG 1 follows Pulsar and ENG 2 follows Pulsar Mod independently per part. See the MIDI guide and registry for custom-controller ranges and encoding.
This sample-based approach adapts pulse repetition, windowing and masking ideas from s3g-dsp’s Pulsar Encoder.