Both lines are cut from one grid. `sync.left` and `sync.right` select sixteen divisions, shortest first: six straight-note major marks with triplet and dotted minor steps between them. Times are the beat times the division. Six-second delay lines preserve literal timing across the full tempo range, including a whole note at 40 BPM.
Feedback and Bounce are two different questions about one number. Feedback is the gain per pass, capped at 0.95 because with the loop filter in the path this module rains rather than self-oscillates. Bounce splits that gain between re-entering the same line and crossing to the other one, and the split is level-preserving, so moving it changes where the tail travels without changing how long it lasts.
Tone is inside the loop, not after it. A second-order high-pass and low-pass sit before each line, so every pass thins and darkens a little more and the repeats recede behind the lead instead of accumulating under it. The 220 Hz default low cut is the interview's own setting, chosen so repeats never mud the mix.
Motion is a sine LFO on each line's read position, the two sides in quadrature, mapping depth onto up to six milliseconds of swing. Because it modulates inside the loop, a repeat swirls further the longer it has been alive - which is the phaser-like behaviour the producer described, rather than a static chorus on the output.
Width and bloom touch the wet path only, so the dry lead's image is never smeared, and Duck is a sidechain from the dry input onto the wet level - up to 18 dB down with a 220 ms release, so repeats step aside while the vocal is singing and come back up in the gaps. Quadrature phase, bloom-after-width order, the duck timing and the feedback ceiling are fixed by design and are not knobs.
The chain stages sit after the dry and wet paths are summed and before the output trim, in one fixed order: reverb, then saturation, then compression. Reverb and saturation are a toggle and a single amount; compression is a toggle with Threshold, Ratio and Makeup, and it runs last because its makeup gain brings the delay feedback up in volume rather than pushing the lead down. Off skips a stage entirely rather than running it at unity, so with all three off the render is sample-for-sample the one the module produced before the chain existed.
Every derivation - seconds, hertz, linear gains, the feedback split - lives in one `expandParams` function in the plugin's schema, which this page, the panel's scope and the compiled kernel all agree on, so a knob means the same thing in a DAW as it does here and as the scope draws it.
