Read or hear → pulse → tap
Sight Read keeps notation central; Echo hides it until you perform from memory. Count-in establishes tempo, and the tap target stays large for touch, keyboard Space, or pointer input.
V0.350 · GAME-FIRST SIGHT READING, ECHO IMITATION, TAP TIMING & EXACT REVIEW
Rhythm Trainer
Choose Sight Read or Echo Imitation, lock to the pulse, and tap with intent. Practice is immediate; Daily and Sprint make repetition easy; fixed A–E Benchmark forms keep local retests repeatable without pretending browser timing is a standardized music exam.
ModePractice
Pattern1 / 5
Meter / tempo—
RunClean
Active time0:00
Practice is ready. Read first; Listen is optional study help.
Pulse Basics
Run paused
The current pattern has not started. Resume when your setup is stable.
Run complete
Mean abs error—
Timing bias—
Within ±60 ms—
Completion—
Consistency—
Weak family—
Family evidence
Recent & best
Exact pattern & onset review
Matched onsets show signed timing error. Misses and extra taps stay separate instead of disappearing into one score.
Complete a run to review it. Echo runs reveal the hidden notation here after you perform from memory.
Benchmark forms, repeatability & trust boundary
Benchmark A–E each contain ten deterministic patterns: two from each family. Listen marks a run Assisted. A local benchmark is comparable only when it is Clean, uninterrupted and at least 80% of expected onsets are matched.
These are browser-local practice metrics. Audio/output latency varies by device, especially over Bluetooth. No percentile, talent score, grade equivalent or standardized-exam claim is made.
Advanced V0.220 Timing & Sight-Reading Evidence Lab
V0.220 DEEP AUTHORITY · LOCAL RHYTHM EVIDENCE
Timing & Sight-Reading Evidence Lab
Train five rhythm families with deterministic progression and benchmark sets. Timing direction, drift, misses, extra taps and family-specific weakness remain visible instead of collapsing into one opaque score.
Choose a Lab run.
Ready
0 / 0 · Load a pattern first.
Last / mean score—
Mean abs error—
Early / late bias—
Timing drift—
Matched—
Complete a trial to see onset-level evidence.
Fixed A–E uses 40 deterministic patterns per set: 8 from each family. Path 50 is 10 levels per family. Custom supports 2/4, 3/4, 4/4, 6/8 and 9/8, 40–200 BPM, 1–4 bars, swing ratio, count-in, metronome and reproducible seed.
Direction, not one opaque score
Mean absolute error shows timing size, bias separates early from late, ±60 ms placement shows tightness, and completion keeps misses visible.
Repeatable by design
Daily is date-stable, Sprint is seed-stable, Benchmark uses frozen A–E forms, and Replay/Share preserve the exact generated session.