Two Beat Rhythm Notes Play the game

Rhythm Notes · Timing and sync

Why Do We Tap Before the Beat? The Science of Rhythm Feel

You hit right on the beat — and the game says early. You switch earbuds and suddenly everything drifts. That happens because the beat a machine measures and the beat a person feels are not the same. Here is what tapping and groove research says, and how a rhythm game lines up sound, picture and judgement.

· Two Beat (GameNow) · 6 min read

We do not react to the beat — we predict it

If you waited to hear a click before moving, you would always be late: just hearing a sound and moving a finger takes about a fifth of a second. So when we clap along, we predict when the next beat will come and launch the movement early. Researchers call this sensorimotor synchronization and have studied it for decades with finger-tapping experiments.

We tap ahead of the beat

Ask people to tap along with clicks and most of them land before the click — untrained tappers typically by a few tens of milliseconds (roughly 20–80 ms), musicians by less. Individuals vary a lot, from almost zero to nearly 100 ms early.

Real drum sounds shrink that lead: in one study people tapped about 20 ms early to a bare click but barely ahead of kick and snare sounds. And when the sound of your own tap arrives late, you start tapping earlier still, because you try to line up what you hear with the beat.

That is why rhythm games need an input offset. Calibration does not fix a “wrong” hand — it measures the habits of a person and a device together. Calibrating against real drums rather than clicks captures the same habit you will have while playing, which is why Two Beat calibrates over a drum loop.

Every sound has its own beat point (P-center)

The moment a sound starts and the moment your ear says “that’s the beat” are not the same. Psychologists call the second one the perceptual center, or P-center. Short, sharp sounds like a kick or snare are heard almost exactly at their onset, but slow-attack or long sounds are heard later; one study found a synth bass with a 33 ms attack was perceived about 20 ms after it started.

Every sound Two Beat judges against — janggu, buk, kick and tap — reaches full level within 2–3 ms. A slow sound like the soft-mallet jing gong is triggered 20 ms early so that its perceived beat lands on time.

Subdividing the beat makes you more precise

When a shaker or hi-hat fills the space between beats, people synchronize more precisely — up to a point; very dense subdivisions help less. Research also suggests that low sounds carry timing best, which may be why ensembles so often let bass and drums keep time while higher instruments decorate.

Two Beat’s Rhythm Playground replaced its click-per-beat metronome with shaker, off-beat chords and bass; in the janggu mode the jing, buk and kkwaenggwari keep the time.

Off-beats and swing: the machine grid vs. the human off-beat

In swing, the off-beat does not fall exactly halfway through the beat — it comes late. Measurements of jazz performance found long-to-short ratios above 3:1 at slow tempos, approaching 1:1 at fast ones. With a 2:1 shuffle at 120 BPM, the off-beat lands 83 ms later than the straight grid. In a game whose judgement window is about ±90 ms, that is enough to mark a perfectly felt off-beat as late — or a miss.

Straight eighths (120 BPM)4 beats × 4 sub-beats · ♩ ≈ 120
둥boom·츠ts·탁tak·츠ts·둥boom·츠ts·탁tak·츠ts·
Straight eighths: boom · ts · | tak · ts · — the “ts” sits exactly halfway through the beat.
2:1 shuffle (120 BPM)4 beats × 3 sub-beats · ♩. ≈ 120
둥boom·츠ts탁tak·츠ts둥boom·츠ts탁tak·츠ts
2:1 shuffle: boom · ts | tak · ts — same 120 BPM, but the “ts” lands two-thirds of the way through the beat.

Korean jangdan divide the beat in three from the start, so this sway is built into the cycle (three per beat: sub-beats). Two Beat judges its shuffle and jangdan stages on that three-per-beat grid.

Groove: a moderate amount of syncopation feels best

Listeners rated rhythms with a moderate amount of syncopation as the most groovy and pleasurable — too square is dull, too syncopated loses the beat (an inverted U). A steady pulse underneath, like a hi-hat, is what lets an off-beat be heard as an off-beat.

So should you randomly nudge notes to sound “human”? When kick or snare was deliberately shifted by 15–25 ms, groove ratings came out lower than or equal to the quantized version; expert players’ real timing rated about the same as the grid, and exaggerated deviations rated lower. Human feel comes from systematic swing, dynamics and ornaments — not random jitter.

That is why Two Beat’s accompaniment is never jittered. Instead, the downbeat is loudest, other beats are about 4 dB softer and in-between notes about 6 dB softer, and only the sixteenth-note shaker leans back slightly, 1.25:1, the way classic hip-hop breaks do.

Eyes are worse at beats than ears

Asked to tap along with flashing lights, people lose sync once flashes come closer together than roughly 0.4 seconds; with sounds they can follow far faster beats. A moving object, like a bouncing ball, does much better than a flash, because you can see its fall and predict the moment it lands.

How much audio–video mismatch we notice depends on direction. The broadcasting standard ITU-R BT.1359 puts the detection threshold at about 45 ms when sound comes first and about 125 ms when sound comes late. When in doubt, it is better to show the picture slightly early.

In Two Beat’s Rhythm Playground a ball hops to each spot you need to hit before you hit it, and the game draws each frame for the moment it will actually appear on screen, about 1.5 frames ahead.

The delays your devices add

WhereHow muchWhat Two Beat does
Audio outputBuilt-in speakers and wired headphones: a few to tens of ms. Bluetooth earbuds: 100–300 ms depending on the modelSubtracts the output latency the browser reports and judges against the beat you are hearing
Audio clockThe browser’s audio clock jumps in steps of several ms each time the device takes audioA clock that follows the tops of those steps, cutting jitter from 2.65 ms to 0.05 ms
ProcessingA loudness compressor delayed every sound by 6 ms (measured)Replaced with a zero-latency soft clipper
InputKeyboards, touch screens and gamepads add a few to tens of msJudges the exact time of the press (event.timeStamp); reads gamepads every 4 ms
Display1–2 frames until a drawn frame is shown, plus monitor delayDraws the song position for when the frame appears, plus a “visual sync” setting

These delays push in different directions: tapping early makes you look early, input lag makes you look late, output lag makes you look later still. In the end, measuring once with your own device and hands is the most accurate fix.

How to calibrate your sync

  1. Use wired headphones or the device speaker if you can. With Bluetooth, always calibrate.
  2. ⚙ Settings → “Calibrate: tap along to the drums”: press any key as the ball lands on each boom and tap. The first four beats are a warm-up and are not counted.
  3. ⚙ Settings → “Match sound and picture”: move visual sync until the ball lands exactly with each drum hit.
  4. After a run, check the timing line on the results screen. If you were consistently off over 20+ hits, a “Sync a → b” button appears — one press applies it.
  5. If you keep seeing only EARLY or only LATE under your judgements, nudge the sync that way.
  6. Recalibrate whenever you change headphones or connect Bluetooth.

What great rhythm games taught us

  • Rhythm Heaven — its creators wanted a game you play by listening, not by watching markers. A perfect hit sounds crisp, a near-miss sounds different, a miss gets a funny reaction — you can play some games with your eyes closed. In Two Beat the sound of your hit changes with your accuracy, and the robot’s rising “whoosh” tells you it is your turn.
  • Taiko no Tatsujin — prints spoken drum syllables (“don”, “ka”) under the notes and uses two-handed big notes and a gauge to build excitement. Two Beat’s gueum line, two-handed “deong” and sinmyeong gauge follow the same idea.
  • Patapon — play well and the music grows into Fever. In the Rhythm Playground, clean rounds in a row add bells to the backing.

Sources

  1. Repp, B. H. (2005). Sensorimotor synchronization: A review of the tapping literature. Psychonomic Bulletin & Review
  2. Repp, B. H., & Su, Y.-H. (2013). Sensorimotor synchronization: A review of recent research (2006–2012). Psychonomic Bulletin & Review
  3. Aschersleben, G., & Prinz, W. (1995). Synchronizing actions with events: The role of sensory information. Perception & Psychophysics
  4. Danielsen, A. et al. (2019). Where is the beat in that note? Effects of attack, duration, and frequency on the perceived timing of musical and quasi-musical sounds. JEP: Human Perception and Performance
  5. Hove, M. J. et al. (2014). Superior time perception for lower musical pitch explains why bass-ranged instruments lay down musical rhythms. PNAS
  6. Witek, M. A. G. et al. (2014). Syncopation, body-movement and pleasure in groove music. PLOS ONE
  7. Frühauf, J., Kopiez, R., & Platz, F. (2013). Music on the timing grid. Musicae Scientiae
  8. Senn, O. et al. (2016). Groove in drum patterns as a function of both rhythmic properties and listeners’ attitudes. Frontiers in Psychology
  9. Friberg, A., & Sundström, A. (2002). Swing ratios and ensemble timing in jazz performance. Music Perception
  10. Frane, A. V. (2017). Swing rhythm in classic drum breaks from hip-hop’s breakbeat canon. Music Perception
  11. Repp, B. H. (2003). Rate limits in sensorimotor synchronization with auditory and visual sequences. Journal of Motor Behavior
  12. Review on synchronizing with moving visual stimuli (PMC)
  13. ITU-R BT.1359-1. Relative timing of sound and vision for broadcasting
  14. W3C Web Audio API — AudioContext.baseLatency and outputLatency
  15. Iwata Asks: Rhythm Heaven Fever (Nintendo)

More notes