Byron — good plan, honest self-critique. Here's what a teacher sees. Short version: you've built a left-hand app. Music lives in the right hand.
#1 — Rhythm and strumming patterns. The single biggest omission. Your app grades where fingers go; it never asks when or how you strum/pick. Beginners who learn from chord charts default to one lazy strum per chord for years. You need: a strumming-pattern library (D–DU–UDU, boom-chick, island strum, travis-pick basics), a pattern assigned to every song, and mic-based grading of strum onsets against the metronome — which is far easier DSP than chord recognition and should ship before chroma matching. Same for piano: LH patterns (whole-note bass → broken chords → alberti).
#2 — Practice structure. You built tools, not a practice. Adults with 20 minutes need the app to run the session: 2-min warm-up → 3-min chord changes → 8 min on one song section → 5 min reviewing old repertoire (spaced repetition — nothing in your plan stops songs from being learned and forgotten). This is the difference between an app that gets used for a week and one used for a year.
#3 — Section looping with count-in. Drilling chord pairs out of context is half the job; the other half is "loop bars 9–12 at 65 bpm until clean." Your tempo-locked autoscroll player can't do this. Highest-leverage player feature.
#4 — Ear training. Lightweight is fine: "is this major or minor," "which of the song's 4 chords is playing," sing-and-match a note against the tuner. Five minutes a week changes a student's trajectory.
#5 — Scales/warm-ups. C major 5-finger pattern on piano, open-position pentatonic on guitar/uke, as the scripted warm-up in #2. Low effort, real payoff.
#6 — Posture micro-content. You can't grade it, but a phone lying flat on a table actively trains hunching. One illustrated checklist per instrument, shown at session start, plus a nudge: "prop the phone at music-stand height."
Guitar: T1: Em, Am, D, A, E · T2: G, C, D7, G7, E7, A7, Cadd9 · T3: Dm, B7, Fmaj7/"mini-F", sus2/sus4 · T4: full F barre, Bm, B, E/A-shape barres.
Ukulele: T1: C, Am, F, G7 · T2: G, C7, A, A7, Dm, Em, D7 · T3: D, E7, Bm, B♭ (first barre) · T4: E, movable shapes up the neck.
Piano: T1: C, F, G, Am root position RH + single-note LH bass · T2: Dm, Em, G7, first inversions for smooth voice-leading (C→G/B→Am) · T3: D, A, E, 7ths, LH broken-chord patterns · T4: all 12 keys, slash chords, sus/add9, hands together.
Public domain (full lyrics OK):
Modern (chords + structure only, no lyrics/melody):
Note how many share G–C–D(7) and C–F–G7: sequence the catalog so each new song adds at most one chord.
Be ruthless about what the glass piano is for: note geography, chord shapes, inversions, and theory visualization — not playing pieces. Practicing repertoire on 14.6 mm keys writes wrong spatial memory that fights his real keyboard. So: the glass shows "here's Am, here are its inversions, here's why C→G/B sounds smooth," then the student plays it on the real instrument; Web MIDI (v1 — correct call) is how the app hears the real keyboard. The falling-notes MIDI view is fine as a reading aid at slow tempo, graded via MIDI input only.
First PD pieces, in order: Ode to Joy (RH five-finger, LH root notes) → When the Saints → Amazing Grace (lead-sheet style: RH melody, LH single bass) → Jingle Bells → Minuet in G (Petzold) → Bach Prelude in C major. Drop Für Elise from the starter list — the A-section is a late-beginner trap that's always played badly; it's a month-6 piece, not week-2.