Time Quantize

Parameters

Strength — How strongly detected transients are pulled toward their target positions, 0–1. 0 = no quantization (input passes through unchanged). 1 = full quantization (transients snap to their targets). Values in between (0.3–0.7) produce humanized quantization where transients move toward the grid but don’t fully reach it — useful for tightening loose performances without making them sound robotic.

Grid Interval Ms — In Var Grid and Combined modes: the time interval (in ms) between grid points, 10–2000. 100 ms = 10 grid points/second; 250 ms = 4 points/second (16th notes at 60 BPM); 500 ms = 8th notes at 60 BPM. Set to your desired rhythmic resolution.

Additional controls

Threshold (dB) — Minimum transient level (in dB) required to be considered for quantization, −60 to 0. Higher values mean only the strongest transients get quantized; lower values catch quieter events too. Adjust this if the node is shifting things you didn’t intend to move (raise threshold) or missing events you wanted moved (lower threshold).

Max Shift Ms — Maximum amount any single transient is allowed to move, in ms, 0–500. Acts as a safety limit — even at full strength transients won’t shift further than this. Set lower (50–100 ms) for subtle correction; higher (300–500 ms) for aggressive rhythmic restructuring. Limits prevent the algorithm from making bizarre jumps when the target position is far from the source.

Mode — Quantization strategy:

• Attractor — transients are pulled toward nearby existing transients (not to a fixed grid). Tightens performance without forcing rhythm onto unmetered material. Best for tightening loose grooves.
• Var Grid — transients are pulled to the nearest point on a fixed time grid set by grid_interval_ms. Classic quantization — forces rhythm onto a beat grid. Best for converting unmetered material into metered.
• Combined — applies both attractor and grid quantization in sequence. Most powerful, most opinionated.

Attraction Radius Ms — In Attractor and Combined modes: how far each transient looks for a “neighbor” to be pulled toward, in ms, 10–500. Smaller radii (10–50 ms) only attract very close neighbors (subtle micro-tightening); larger radii (200–500 ms) can pull distant transients together (aggressive rhythmic snapping).

Attraction Falloff — In Attractor and Combined modes: how the attraction strength varies with distance:

• Linear — strength falls off proportionally with distance. Predictable, even attraction.
• Expo — exponential falloff; very strong nearby, very weak far away. Use for “snap to closest only” feels.
• Gaussian — bell-shaped falloff; smooth, natural-feeling attraction with a defined “preferred distance.”

Phase Strategy — How the spectral data is shifted in time when transients move. This affects sound quality, not which events are quantized:

• Direct — direct complex copy of frames; cleanest for short shifts but can click on long ones.
• Source — copies from the source position with phase preservation; balanced quality.
• OLA — Overlap-Add; uses windowed crossfading between source and target. Smoothest for medium shifts.
• Vocoder — phase-vocoder reconstruction; highest quality on tonal/sustained material but most CPU.

For most rhythmic material, OLA gives the best results.

Crossfade Frames — In Direct and OLA phase strategies: number of STFT frames used for crossfading between source and target positions, 1–16. Higher values produce smoother shifts but slightly blur the transient. 4 is a balanced default.

About Time Quantize

Time Quantize is a creative spectral-domain quantizer with three strategies. Unlike DAW quantize (which moves note onsets in MIDI), this works on detected audio transients. The Attractor mode is great for tightening performances — pulling close transients together — without forcing rigid grid placement. Var Grid and Combined modes are for harder rhythmic correction. The phase strategy options matter especially on long shifts: Direct is fast but can artifact; OLA and Vocoder preserve sustained material more cleanly. For percussive material with tight quantization, Direct is often fine; for shifting through sustained tonal content, Vocoder is worth the extra CPU.

Generated 2026-05-05 from K2K_Dev@96730bdc by scripts/gen_lexique.py. Edit _intros/ or _overrides/, not this file.