Can You Copy a Swing Setting Between a DAW and a Drum Machine? A Timing Audit

The control label does not define a cross-product conversion

Official manuals already show why a bare number is incomplete. Logic documents six 1/16 Swing quantize choices—50, 54, 58, 62, 66, and 71 percent—and says every second grid point is delayed. MPC 3.9 exposes Swing from 50 to 75 alongside separate Time Division, Shift Timing, Window, Strength, and event-scope controls. Yamaha SEQTRAK instead describes the timing of even steps as −58 to +58 ticks, with one step equal to 120 ticks. Roland P-6 uses a −90 to 90 Shuffle range and calls 0 no shuffle. Ableton Groove Pool separates Base, straight Quantize, Timing, Random, Velocity, and Global Amount.

These manuals do not prove that two identically named percentages must differ. They prove a narrower point: the display alone is insufficient evidence of equivalence. Measure the actual events under declared conditions.

Build a straight probe before turning Swing on

Create a new one-bar 4/4 clip plus the first onset of the next bar. Use one short sound, one note number, one fixed velocity, one fixed note length, and the same tempo in source and target. Place sixteen notes exactly on a sixteenth-note grid. Disable groove, swing, random/humanize, velocity transformation, note chance, delay, arpeggiation, and live input. Save this untouched version as S0.

Here each A–B–C unit spans two sixteenths, so A and C are neighboring eighth-note grid points and B begins halfway between them in S0. If you test eighth-note swing instead, declare an eighth-note base and let each pair span one quarter note. Never compare a source’s sixteenth swing with a target’s eighth swing merely because both panels show the same number.

1. Freeze the source: duplicate S0 to SX; record product/version, tempo, meter, grid or Base, Swing/Shuffle value, strength or amount, window/range, event scope, track/clip scope, and global amount.
2. Remove unrelated motion: set Random/Humanize and velocity-groove influence to zero for the timing audit. Keep one sound and a stable monitoring path.
3. Make playback timing inspectable: on a duplicate, use a documented event-preserving Commit or destructive apply, MIDI export, MIDI recording, or equivalent route that still exposes MIDI note starts. Name the result E. Preserve S0 and SX; Logic documents Region Quantize as non-destructive, while Ableton documents that Commit moves MIDI notes. An audio render needs a separately declared transient-measurement method and is outside this MIDI-event audit.
4. Verify the control: compare S0 and E. If no intended second onset moves, stop. The effect may be off, aimed at another resolution/scope, excluded by a window, or still hidden in playback state.

Use A–B–C as a neutral coordinate, not a new vendor standard

For each pair, read the event editor’s musical position or the finest documented timing display available. Subtract within the same unit:

The normalization makes different tick resolutions comparable without pretending that their knobs share a scale. It is deliberately incomplete: translating A/B/C together or scaling the whole A–C span leaves p unchanged. Convert all three events to one musical coordinate first, then compare the absolute A and C positions and C − A as well as p. If a global phase offset is intentionally allowed, declare it before the test, subtract that one offset from every target position, and report phase transfer separately.

Declare the target’s smallest event increment and the direction used for an exact half-increment tie before transfer. Define signed error as target − desired source position in the converted target unit: positive is later, negative is earlier. Classify the result as exact only when A/B/C all match; as a nearest-representable approximation when A and C plus the span match and each B is the nearest target position with an absolute error no greater than half that increment; otherwise classify it as failed. Record every signed error, and never call an approximation exact. Example: if A = 0 ticks, C = 24 ticks, and the converted source B = 15.6 ticks, source p = 65%. A whole-tick target places B at 16, so target p = 66.7% with a +0.4-tick error—an approximation, not an exact transfer.

Measure every pair, not only the first. A groove template can vary from pair to pair. A window can exclude some events. Ableton documents that Random can offset every voice differently; that behavior must remain off during a deterministic timing transfer, then be tested later as its own version if desired.

Choose one of two transfer paths—do not stack them accidentally

Committed event transfer asks whether explicit note positions survive the handoff. Recipe transfer asks whether the target can generate the same timing relationship from a straight grid. If E already contains shifted second notes, applying target swing to T0 begins a second processing stage. Depending on the target’s Base, quantize behavior, strength, and eligibility rules, positions may stay put, snap elsewhere, replace the prior offset, or move again. Test and label that stage separately; it cannot diagnose the one-stage import.

After A/C positions, pair spans, and p values match, create a separate musical comparison. Restore the intended samples and velocities without changing note starts. That combination reproduces the measured onset relationship; it does not reproduce sample envelopes, accents, bus processing, playback latency, or every changing detail of a groove template.

Audit all eight pairs, simultaneous hits, and the loop seam

Copy this ledger before auditioning. Use exact displayed positions or normalized values; do not fill it with estimated listening impressions.

Then add one kick exactly on an A and one snare exactly with a B, on a duplicate only. Repeat the transfer. The kick tests whether an intended anchor stays fixed; the snare tests whether simultaneous voices remain simultaneous. If target randomization separates two events that shared one onset, you have changed inter-voice alignment as well as timing. Finally, loop at least four passes and inspect pair 8 into bar 2: record C as the unwrapped bar-2 A (step 17), even if a loop display wraps it to zero. An export boundary, clip crop, or target loop setting must not erase the final B or move that C.

Let the first repeatable mismatch choose the next move

Swing-setting transfer questions

Is 66.7% the “correct” Swing amount?

No. In this audit it is only the location of B at two thirds of an A–C unit—the 2:1 long–short triplet landmark. Music can use straighter, earlier, later, changing, voice-specific, or performance-derived timing. The source material and musical goal decide; the formula only documents one relationship.

Can I copy 60% from one device to another?

You can try it as a starting hypothesis, but the shared display value does not prove shared output. Keep the same musical subdivision and eligible event set; record and freeze each product’s local Strength/Amount, Window/Range, and scope independently without assuming equal display values. Then inspect every A/C position, pair span, p value, and the unwrapped seam.

Should I import swung notes or recreate Swing on the target?

Choose based on what must remain editable. Importing committed events preserves explicit positions if the handoff passes. Rebuilding from a straight copy keeps a target-native recipe. Keep those paths separate, and never interpret a double-processed clip as a failed one-stage transfer.

Why not judge only by ear?

Listening is the final musical gate, but it cannot identify whether a difference came from note position, velocity, sample envelope, effect, or latency. The ledger first isolates the timing component; the later listening pass judges the complete result.

Does matching one hat pattern reproduce a whole groove?

No. It verifies the tested pair positions under frozen conditions. A full groove may move different voices differently and may include velocity, note length, random variation, or multi-bar structure. Add those one at a time after the timing baseline passes.

Which of these timing controls does FingerDrum provide?

FingerDrum has its own project-level Swing amount from 0–75% with a 1/16 or 1/8 grid. It shifts eligible sequenced drum and melody events. Changing Swing does not move recorded layers by itself; post-record quantize can align a recorded layer to the current swing grid. This does not claim that FingerDrum implements Ableton’s Groove Pool or MPC’s Timing Correct controls.