Spacebar Rhythm Game

Easy, Normal, and Hard change generated beat spacing rather than loading a commercial song. Sixteen targets form one round. Press Space near the target: Perfect awards 100 points, Good awards 60, and Miss awards zero. Perfect and Good extend combo; Miss breaks it. Restart creates a fresh local round.

Perfect covers offsets from −70 to +70 ms. Good covers the remaining offsets through ±150 ms. Anything outside that window is Miss; a target also becomes Miss when no input arrives by 220 ms after it. Negative offset means early and positive offset means late. Accuracy weights Perfect as 1, Good as 0.6, and Miss as 0.

Sound uses a short Web Audio oscillator tone generated at runtime after a user gesture. No recording, sample, melody, or third-party audio asset is included. Visual-only disables the AudioContext path. The animation loop uses requestAnimationFrame and the judgement clock uses performance.now rather than counting setTimeout ticks.

Judgement is written as text and is not communicated by color alone. The target also changes shape and position treatment, while reduced-motion CSS removes unnecessary transitions. Players can keep visual cues with sound off. The design avoids rapid flashing and does not place advertising or navigation controls inside the play field.

Audio output, screen refresh, browser scheduling, operating-system input, and keyboard scanning can shift offsets. Average offset helps identify a repeated early or late tendency but is not an automatic calibration claim. Scores stay on the page, contain no typed text, and are not mixed into the normal CPS leaderboard.

Look at the sign and spread of several offsets rather than one judgement. Mostly positive values indicate late input along the combined display, perception, and keyboard path; mostly negative values indicate early presses. Start on Easy, move to Normal when cues remain readable, and use Hard only when its denser schedule stays comfortable. Difficulty changes target spacing, not judgement windows. Rhythm results do not establish music skill, esports rank, or professional ability.

Accuracy divides the sum of documented judgement weights by all scheduled targets, so a missed target lowers the result even when no key was pressed. Score and accuracy therefore answer different questions: score uses 100 and 60 point values, while accuracy expresses weighted completion as a percentage. Combo shows an uninterrupted sequence of Perfect or Good judgements, and max combo preserves the longest such sequence. Pausing or hiding the page ends the active round because resuming across an unknown scheduling gap would make the beat timeline misleading.

Browser audio commonly remains locked until the player presses Start, which is why the game does not attempt autoplay. The generated oscillator is a timing cue, not a musical recording. Sound-off and visual-only choices remain valid ways to play, and neither changes the judgement mathematics. If a display or wireless keyboard adds a stable delay, the average offset may move while the player’s relative consistency looks similar. Compare rounds on the same setup before changing any advanced latency assumption.