Learn Piano the Right Way

Every take answers two questions, not one

Sit down and play a scale. There are two entirely separate things you might want to know afterwards, and it is easy to spend years collecting the answer to the first one while assuming it covers the second.

1. Did I play the right notes? A closed question with a yes or no answer, checkable instantly, by a person or a machine.
2. Did I play them well? An open question about how even the notes were, whether the volume stayed where you meant it, how long each note actually sounded, and whether both hands were equally steady.

The right notes are the floor, not the goal. A scale can be completely correct and still be lumpy, patchy in volume, and clipped short in the second octave. Every note "right", and it still does not sound like someone who can play. The gap between those two questions is where the years go.

What piano apps are genuinely good at

If you have tried learning with an app, you have probably met the falling-note screen: notes descend towards a keyboard, you press the matching key as they land, and the app marks you as you go. It is a good design, and it is worth being precise about the real problems it solves.

• It gets you playing immediately. No teacher, no waiting until you can read music, no first three weeks of theory before a note is heard. That is a genuinely hard problem and it is well solved.
• Note reading becomes automatic. Linking a symbol on a stave to a key under your finger is pure repetition with instant correction, which is exactly the kind of task software is better at than a human being.
• It gives you a reason to come back. Repertoire you want to play, a streak, a sense of progress. Say what you like about gamification: practice you actually do beats a better method you abandoned in March.
• It keeps you moving. Playing along with something in tempo trains you out of stopping dead to fix every mistake, which is one of the most stubborn habits in amateur playing.

The question note matching cannot answer

Matching notes asks, for each note in turn: was that the right key, near enough to the right moment? It is a pass or fail, note by note. That is what makes it fast, forgiving and encouraging - and it is also what puts the second question permanently out of reach.

Two people can play the same scale, both hit every correct key inside the window, and produce performances that sound nothing alike.

A tick tells you that you did not fail. It does not tell you what to do tomorrow. And between lessons, "what should I do tomorrow" is the entire problem - which is why the most common outcome of practising alone is not getting worse, it is getting stuck, politely, for months.

What real feedback has to do

Useful feedback names something you can change before the next take. That sets a fairly demanding bar: it has to measure how you played, separate the measurements that fail for different reasons, and point at a specific note.

Here that means four numbers, reported for each hand on its own, from one take on a MIDI keyboard.

Scoring each hand separately is the part that changes what you practise, because the two hands do not stumble in the same place. In C major going up, the right hand passes its thumb under between the third and fourth notes, E to F. The left hand crosses its third finger over between the fifth and sixth, G to A. Two different events, at two different moments, in a scale most people practise as one thing.

Under the scores is a short list headed What to work on, which names the fault and the note it happened at. That is the difference between a number and an instruction: "Timing 84" tells you to practise more, and practise more is not feedback. "Dragged at the fourth note, ascending" tells you what to practise and how you will know when it is fixed.

What none of this can see

Worth stating plainly, because a tool that claims everything is not worth trusting on anything. A MIDI keyboard reports which key went down, when, how fast, and when it came back up. Every measurement above is built from those four facts, so none of the following is visible and none of it is claimed:

• Your hands. Posture, wrist height, curved or flat fingers, tension in the forearm. A keyboard has no idea where your body is.
• Your fingering. It cannot tell which finger played a note, so it cannot tell you that you used the wrong one. It can only tell you the result was uneven.
• Tone. On an acoustic piano, tone is a real thing produced by how the key is moved. A velocity number is one measurement of speed, not a description of sound.
• Musicality. Whether the playing was shaped or expressive. What is measured is evenness, which is a different thing - and mostly the thing scales are for.

Dynamics and note length also need a real MIDI keyboard: they are built from how hard and how long each key was held, so they are unavailable if you are playing on the computer keys.

So what does the right way look like?

Unglamorously, it is five habits, and none of them require an app:

• Play slowly enough to be accurate, then raise the tempo in small steps. Speed practised badly is speed learned badly.
• Separate the hands when something breaks. Find out which hand is actually the problem before you spend a week on both.
• Fix one named thing per session. One fault, one place, checked at the end. Not "work on scales".
• Decide the order before you sit down, so a tired evening does not turn into playing what you are already good at.
• Measure something. Otherwise "better" is a feeling, and feelings about your own playing are famously unreliable on the day.

An app earns its place if it makes the last two easier. The rest is you, at the piano, paying attention.

Go deeper

• How grading works What a take actually measures, note by note, and what a MIDI keyboard cannot see.
• How to practise scales evenly The tempo ladder and what to listen for at the thumb turn.
• A routine that survives a bad day What to put in a session, in what order, and why the order matters most.
• Hands separately or together? Why the two hands stumble at different points, and how to find out which one is the problem.