A plan is optional. The log, metronome, recorder and timing analysis all work without one — this app is a practice tracker first. A plan just decides what goes into each of the four areas each week, so you don't have to.
If you want one: tell it what's actually going wrong, answer a few questions, and get a week-by-week structure that plugs into the Log tab. Week 1 is always diagnostic and conservative; the plan is revised from what you actually record. You can drop it any time.
Two alternatives. You can write your own plan by hand — tap the Guide tab, follow the daily session and the BPM ladder, and set the Focus banner yourself; the app works fully without a generated plan. Or, if you use Claude or another assistant, copy the prompt below, paste it in, and paste the JSON reply back here.
Save settings changes what future adaptations use — weeks already written stay exactly as they are, and nothing is sent anywhere. Rebuild remaining weeks rewrites this week onward using the new settings; weeks you have already been through are kept, and your log is never touched.
A practice tracker. Log what you did, record yourself, and see whether your tempo and steadiness are actually improving over weeks — that is the whole point, and it works on its own.
The plan is an optional layer on top. If you want one it suggests what to put in each area each week; if you don't, set the Focus banner yourself and use the four areas below as the shape of a session. Nothing in the app requires a plan.
Tap the 📅 date above the week strip for a full month calendar (gold = practised, red dot = recording, hatched = rest day). The ‹ › arrows step one day; tapping a key on the strip jumps to that day; tapping the already-selected key opens the calendar too. You can log any past day, never a future one.
Open this tracker first, before touching the keys. Start the metronome, then play. The tracker being open is the cue; practising is the habit.
Right notes, right fingering, no hesitation, no rushing or dragging against the click. A wrong finger with the right note is not clean — it will fall apart at speed. Log the fastest tempo that meets all four, and answer Notes clean? honestly under steadiness — the timing analysis can't hear wrong notes, so that tap is what tells the ladder rule and the plan whether a tempo really counts.
Before tempo work, check: sitting tall with forearms level to the keys, wrists neither collapsed nor raised, fingers curved, shoulders down. If your forearms tighten or ache when you push tempo, that is your body vetoing the BPM — drop 10, shake out, and log “Tension / tired hands.” Speed built on tension does not last and can injure. A minute of slow, loose playing at the end of every session is not wasted.
Start at a tempo where you are certain you can play cleanly.
Never negotiate with the rule. It removes guesswork about when to push and makes the chart in Stats mean something.
Use the problem bars line under each piece to name the exact bars that break — then loop only those (One Bar Loop) before running the whole piece.
Quarter for pieces. Eighths for your subdividing drill — the light off-beat click is the "and" in 1-and-2-and; count with it out loud. Triplets when they show up in Alfred.
Gap mode plays one bar with the click and one bar in silence (or 2 and 2). If you're still with the click when it comes back, your internal pulse is real. This is the true test of steady tempo — use it once your piece is comfortable.
TAP: tap along with any tempo and it reads the BPM. Vibrate works on Android for silent practice. The metronome keeps running when you switch tabs.
Record the final play-through every session — the button is right above the steadiness rating on purpose. Listen back before you rate. Once a week, use ▶ Last week's to hear the same piece seven days ago. That comparison is where progress becomes audible.
Inside the Claude app the microphone is blocked by the app's sandbox — that's not a permission you can grant. Two ways that always work:
After recording, tap Analyse timing. It finds every note attack in the audio, lines them up against an eighth-note grid at your target BPM (or detects the tempo if you leave it blank), and reports how far off the grid you were, what percentage landed within 25 ms, whether you rushed or dragged, and the tempo change from start to end. Dots above the line are late, below are early; red dots are the outliers.
It gives a suggested steadiness score you can accept with one tap. It measures timing only — it can't hear wrong notes or fingering. Best reads: play a single-line passage or a clear-attack piece, metronome on headphones (or off), phone close to the piano.
Your streak survives one rest day per week. Two missed days in a row, or two in the same seven, break it. Best streak is kept forever, so a broken chain never erases what you did.
Describe what's going wrong in a paragraph, answer five questions, and get a 4–8 week plan. It adds the plan's pieces and chords to your tracker, sets the Focus banner to the current week, and rewrites the four skill-area lines in Log with that week's drills. Each week has a checkpoint you verify by recording.
Adapt next week from my log is the point of the whole thing: after a few sessions, it reads your BPMs, steadiness, notes and timing analyses and revises next week — advancing if you met the checkpoint steadily, changing the drill (not repeating it) if you didn't, putting a skipped area first, adding gap-mode work if you rushed. Do it every Sunday as part of the weekly ritual.
Without a plan the app still works — the four areas, the ladder rule and the analysis don't need one. The plan just decides what goes in each area each week.