Resonant Octave Mist — complete Csound settings
Note and effect connections #
Instrument 1 uses p3 for held seconds, p4 for MIDI pitch and p5 for velocity from 0 to 1. Its left and right outputs accumulate on shared buses. Instrument 90 processes the buses once per control block, clears the inputs and keeps effect memories active through the tails. Envelopes with release segments extend the note automatically. The score is an audible pitch-and-touch example; the synthesis controls are independent of its particular notes.
Csound interpretation #
These are explicit Csound models, not identical implementations of the Ableton devices. Wavetable positions, proprietary oscillator engines, physical-model macros, filter circuits, envelope slopes and effect algorithms can differ. Modulation depths are specified in octaves, cents or milliseconds rather than treated as universally interchangeable matrix values. Controller inputs such as pitch bend, modulation wheel, pressure and per-note expression are not mapped. Independent score notes do not implement Live's monophonic note-priority or legato system. Any pitch memory or drift included in this instrument is described in its steps.
Signal path and controls #
Stage 1 — Make a narrow, singing noise band #
Filter quiet white noise through two narrow resonances. The center is 890 Hz at MIDI 69 and tracks the keyboard; the channels differ by seven cents. A 0.30 Hz filter motion and slower pitch drift loosen the ringing. Shape amplitude with a 2.20 s attack, full sustain and 3.20 s release. This noise-band model approximates the source resonant tone without implementing Drift's self-oscillating filter.
| Csound control | Value |
|---|---|
| Noise / instrument | −24 / −12 dB |
| Centers | 890 Hz at MIDI 69, full tracking; 7-cent channel separation |
| Bandwidth | Center / 60 |
| Motion | 0.30 Hz, ±0.003 octaves; slower ±3 cents |
| Amplitude | 2.20 s attack, full sustain, 3.20 s release |
Stage 2 — Add the upper octave echo #
Use two overlapping 100 ms delay windows to shift the preceding signal by a frequency ratio of 2, plus 3 cents. The delay begins at 375 ms and moves across the additional window. Blend 4% shifted signal, lowpassed at 2000 Hz, with 3.3% local feedback. The next stage receives this whole blend, so series octave relationships also interact. This moving-window shifter differs from Ableton Shifter and adds window latency.
| Csound control | Value |
|---|---|
| Pitch ratio / cents | 2 / 3 |
| Base delay / window | 375 / 100 ms |
| Feedback / wet | 3.3% / 4% |
| Tone | 2000 Hz |
Stage 3 — Add the lower octave echo #
Use two overlapping 100 ms delay windows to shift the preceding signal by a frequency ratio of 0.5, plus -3 cents. The delay begins at 500 ms and moves across the additional window. Blend 2.5% shifted signal, lowpassed at 1800 Hz, with 6.8% local feedback. The next stage receives this whole blend, so series octave relationships also interact. This moving-window shifter differs from Ableton Shifter and adds window latency.
| Csound control | Value |
|---|---|
| Pitch ratio / cents | 0.5 / -3 |
| Base delay / window | 500 / 100 ms |
| Feedback / wet | 6.8% / 2.5% |
| Tone | 1800 Hz |
Stage 4 — Surround the tone with a damped room #
Feed the preceding signal through 10 ms predelay into reverbsc, with feedback 0.9383, 1800 Hz damping and 50% wet mix. Predelay leaves the initial articulation in front; feedback prolongs the tail, while damping darkens repeated circulation. The source room decay setting is 11 seconds, but Csound feedback is not a decay-time control: this is an interpretation of that space, with different diffusion and modulation.
| Csound control | Value |
|---|---|
| Source decay target | 11 s (not a measured Csound decay) |
| Predelay / feedback | 10 ms / 0.9383 |
| Damping / wet | 1800 Hz / 50% |
| Room side multiplier | 1 |
| Room input cuts | Off |
Output and signal rates #
The script uses 48000 Hz, 16-sample control blocks and stereo output. The fixed output-scale control is 216.0613 and follows the whole chain at every stage, separate from any final gain step. DC-removal blocks, where enabled, use order 2048 to limit low-note attenuation; they add processing delay.