Piano World Home Page Forums Our Most Popular Forums Adult Beginners Forum Minuet in G (the other one) by Spaced Repetition System

There’s been some discussion of spaced repetition systems (SRS) on PW recently. For example:
Piano Practice App with Leitner, SRS & Interleaved Learning?
Applying Spaced Repetition to Piano Practice?

I’ve been very interested in SRS recently, and am very happy with my use of it in other areas of learning. So, as an experiment, I’m going to try memorizing Minuet in G (BWV Anhang 116 from the Anna Magdalena Bach Notebook). This is the “other” Minuet in G, that starts with a G major arpeggio, not the famous one. (I’m also currently memorizing the famous one, but not with a formal SRS schedule.)

My aim is to set out observations and learnings along the way as a result of using the SRS formal plan. I hope this will be a way for me and others to observe how an SRS might work with one aspect of piano practice (memorizing), and generate ideas about what kinds of adjustments or flexibility might be useful to adapt the insights behind SRS to piano practice.

I’ll put my thoughts about the logistics of preparing for this in a separate post.

I’ll also write a post about what SRS means to me, to make this thread self-contained and also to highlight the features that I particularly want to be sure to use in this project.

Very cool! I'll be curious to see the results of your experiment!

Most of the common strategies that teachers recommend for learning a piece are already different forms of “spaced repetition”. So keep that in mind, you don’t have to re-invent the wheel. But understanding the idea can help fine tune your practice strategy. For me it’s mostly about ”more effectively’ scheduling the bits and pieces of my practice.

Also, SR doesn’t completely replace standard repetitions, that’s still part of practice at some level.

Re: Apps (Anki, SuperMemo) I’ve experimented with these two apps and found they weren’t well suited for Piano Practice. After trying for about a week with SuperMemo, I just went back to my manual system (box of index cards). The Music\Piano Practice Assistant app looks promising, but haven’t had a chance to try it out yet.

The one app that I did find useful is called Flashcards+ by Chegg. I create a deck for each project I work on. So for each piece of repertoire - I have a deck. For each chapter in my books - I have a deck. And I have several decks for practicing different technical exercises.

So it’s easy to interleave my practice by just turning on the randomizer. I can also shuffle decks together and interleave them that way. Plus it’s a whiz at keeping me organized. If I get interrupted in the middle of practice it keeps track of what’s completed and what still needs to be practiced.

One neato feature - the app lets you take photos with your phone/tablet and put them on the cards. So many of my cards don’t have any text at all, just photos of the notation that it’s prompting me to practice.

Good luck on your experiment, I’m interested to see how it turns out!

I will be watching with interest.

Spaced repetition does work. Sometimes we practiced a piece for 2 hours and don't get very far. But after taking a break for a few hours you start seeing the same notes from a different angle and all of a sudden you are able to put the piece together.

Besides just practicing a piece for 2h and then come back later for another 2, I'd normally start with a slow practice to learn the notes properly. Break the piece into manageable chunks like the top half for the first hour sort of thing. And before ending a practice session usually take a recording to track my progress. If I decide not to practice any more on any given day, I can still listen to the soundtrack to learn the tune (passive learning). And the places you feel you may have wrong notes, you need to work on them individually and tie them back to the main piece.

Once someone in the family went for a Gr. 1 level Conservatory exam for violin playing "Minuet in G". All she needed to read was the Treble Clef. 2 weeks before the exam started recording and ended up passing the exam barely. Don't think she got any fewer than 5 mistakes after 2 weeks of practice.

Would be interesting to see how far you can go from the beginning when you're sight-reading a new piece to when you get to play the piece note perfect.

thepianoplayer214, experimenting with a piece-from-scratch in the future would be interesting. For this piece, I can already play it from the score so mostly the only thing to be learned here is the memorization. Eventually perhaps I’ll also find out if I can relate musically in a different way when I’m playing from memory vs. when I’m playing from the score. But that is something for after this SRS memorization project.

I have come to find that 2 hours, or even 1 hour, or perhaps even 30 minutes, is too long on a piece for me. I slog away thinking I’m making progress and then the next day it has all dissolved. One thing that my experience with SRS encourages me to do is change that completely and just work on very very small chunks of new material each day: on this Minuet, 4 measures with one hand HS, or 4 measures HT after the HS are learned. (Others might use SRS with a great deal more new material each day; the point is, that for me SRS is an avenue into understanding “small chunks” in a way I never did before. Others might grasp “small chunks” a lot faster than I have of course, and without needing SRS to hammer the idea into their head.)

You remind me though that it’s probably helpful for me to try to make clear the principles I’m finding as I work through this system. Judged on the surface details, it can seem like it’s just about bookkeeping, but at a deep level for me it is a way to put principles into action.

I think this is wise. I was surprised to read the reference to 1 and even 2 hours. Maybe that's why the student who did the Minuet as grade 1 violin piece made so many mistakes. One doesn't gradually make less mistakes with proper practice - the mistakes don't come in, in the first place. The shorter practice of less material you're talking about would tend to lead to that.

In my head, I supposedly know as facts the principles of effective practice. But I've always had a really hard time putting them into effect; it's like they're in my head only and don't reach my gut.

I've recently had a highly successful experience with SRS in a different area, and as I think about how to use it in other learning areas, it's like it's turning on all my internal lights for understanding principles of learning in a gut way. It also, by contrast, has me thinking about areas of learning that are perhaps not really adapted to an SRS. So for example, maybe learning the fingering for a scale is a good SRS task, but learning to perform the cross-overs smoothly is not a good SRS task.

But fundamentally, it seems to have turned the light on for me of: "small bits, every day".


Yes, and that seems to be part of the thing I mentioned above, about some things that aren't suitable for SRS.

For example the initial learning of 4 bars of my minuet, yes, repeat a few times, get it into short-term memory, OK good, let it lie until the next SRS repeat, and then if I remember it, let the time between repeats grow.

Whereas, figuring out how to do a good thumb-under is something that I think needs some work every day -- it will take me a lot more than a few repeats to make the movements fluid and ingrained in my body. And while I'm working on it, I probably need to work on it every day, not subject it to the increasingly-spaced repeats of SRS.

Another example of this is longtones warmups on my flute: that's never going into an SRS system for me: it's part of every day getting ready to play, checking in, and continuously improving my technique. Yet even here the success I've experienced with SRS in another area has helped me to grasp an enormously important idea for music (or any area): small bits of work, done consistently over time, add up.

Intervals don't have to increase, they can be consistent, for example, if you're working on something every day at 10am, that is still 'spaced repetition'.

The two ideas that seem like a good guide to creating spaced reps:
1. Anything that clears out your 'working memory', like taking a break or working on something else for a bit.
2. Getting decent sleep in-between the repetitions.

Hmm, so I'll try to describe spaced repetition, and spaced repetition systems (SRS).

-- repetition
Spaced repetition is based on the idea that when you retrieve something from memory, you also strengthen the memory. So to learn something, you want to practice not just learning it the first time, but also practice retrieving it from memory. Each time you retrieve it from memory, this strengthens your memory of it such that you can go a longer amount of time without forgetting it.

-- spaced repetition
Because you're building up the length of time you can remember the item, that means you can leave longer and longer spaces in between your retrievals. You can use this property of memory to set up a schedule of retrievals at longer and longer time intervals to get to longterm (essentially infinite) memory while also reducing the total amount of time you need to spend practicing the retrieval.

-- system
A spaced repetition system is any system that helps you keep track of the items you want to practice retrieving, and when you want to practice them. That can be done informally and mentally, or slightly more formally on paper, or extremely formally with a computer program.

-- small bits
One of the principles for entering items into it is that you should be working with small units of information at first, and then as you learn the small bits slowly combining them into larger bits. So for example, learn 4 measures. Then learn 4 different measures. Then practice joining them together into 8 measures. This is better than trying to learn 8 measures all together as the very first task.

(Choosing the smallest unit for the bits will depend on topic and person. Other people might be able to learn 8 measures at a time; or on other pieces I might not even be able to manage 4 measures at a time.)

-- patience and consistency
Learning a piece incrementally is, as Groove On points out, is a well-known principle. But somehow it's only with my success with SRS in a totally different area (a programming language, to be specific) that I can really envision myself having the patience to put this into effect with memorizing a piece. I believe that my success with the patient incremental approach in an SRS has also increased my ability to take a patient incremental approach in non-SRS tasks -- for example, my daily longtone warmups on my flute.


[ETA: cross-posted with Groove On: my "non-SRS tasks" would be his "intervals don't have to increase" tasks, I think. I may change my language to Groove On's language for future posts.]

I've setup my system and will start tonight.

-- tl;dr summary
Decide on what you'll study, and in what order ("units"). Then setup a system for scheduling the units ("mechanics").


Details follow for those who want to delve deeper:

-- units
I'm going to proceed 4 measures at a time from the end of the piece, periodically gluing them together into longer units.

I'll learn each 4 measure unit first LH, then RH, then HT. I usually learn HS starting with RH first, but I'm experimenting with prioritizing the LH so perhaps it won't feel like some mysterious add-on underneath the better-known melody.

When I learn a 4 measure unit, I'll also include the first beat of the following measure. That should reduce the work necessary for the gluings without overloading the pre-gluing tasks.

The form of the piece is [A A'] [B C A''], with each bracketed part repeated. I've labeled the cards with the letter of the phrase to be learned, and which half. So for example, for A'', I have cards for 2nd 1/2, 1st 1/2, and All, and within each of those, cards for LH, RH, and HT. That makes 9 cards total to learn the 8 measure unit A''. Then 9 cards to learn C. Then 3 cards to learn CA'' (all LH, all RH, all HT). And so on.

The sections and gluings that I'm using are:

A''
C
C A''
B
B C
A'
B C A''
A' B
A
A A'
B C A'' B C A''
A A' A A'
A A' A A' B C A'' B C A''

Plus one card at the start for the definition of the form itself (that is, the abbreviation [A A'] [B C A'']. That makes 70 cards.


-- mechanics

I've written each of my 70 units on a card and put them in the order I plan to learn them.

I'm using the schedule from the book Fluent Forever, which is a 64-day cycle. So my filebox is set up with dividers from 1 to 64. These mark decks of cards. All cards start in deck 1.

Each day I try to play from memory all the cards from all the scheduled decks for that day, apart from deck 1. These are "review" cards. If I successfully remember how to play a card, then it goes into the next numbered deck. If I can't remember how to play a card, it goes to the front of deck 1. The card will come up again for memory practice the next time its new deck is scheduled.

I also study from deck 1 every day, but only a limited number of cards each day. These are generally "new" cards (plus cards that I learned before but then forgot and have started over on learning, but I'm going to call them all "new" cards).

Because I feel very unconfident of my ability to memorize music, I'm going to generally only do 1 or 2 new cards a day. By contrast, in my programming language SRS, I'm doing 5 new cards a day. I've seen up to 20 new cards a day suggested, depending on the learner and the time they have available.

Usually in SRS, the front of the card has the question or prompt, and the back has the answer. For this Minuet project, all the backs are blank because I'm going to look up the answers in the score if I can't remember them (and of course when I'm first learning them).

A lot of pieces we play (even if we don't know the tune that well) somebody already posted a soundtrack online. We can just listen and learn the tune.

Does it matter if you read as you play or you do it from memory? I've seen professional "concert" pianists who play some pieces from memory and others with the score. As long as you can show enough expression in your playing. A while ago came across somebody who posted Pop songs online like 3/wk. After a while started getting bored with his performances like every pieces has no dynamic variations (staying in the middle loud mf range)... and his phrasing looked robotic.

A lot of Pop songs tend to be simple (staying on the same Key for the whole piece). For someone who learned to play 3 per week he didn't try to memorize any of his pieces, just being a good sight-reader and get the pieces online quickly.

A year back heard a Christmas tune on radio. Didn't get around to it until this year "In the Bleak Midwinter" by Gustav Holst. The piece has only 4 lines. The first day was sight-reading. The next day was to get comfortable playing it. Church hymns tend to be repetitive. Don't think I'm playing it the first day with a good flow instead of slowing down every few minutes. Giving a time gap between practice sessions is necessary.

I think one of the challenges is that the optimal 'Pattern of Repetition' for learning changes over time. Especially when you consider we're not just trying to "memorize' facts, we're also 'chunking' things together into higher level skills. So the question is - which 'Patterns of Repetition' are optimal in the different stages of learning?

In the beginning, when I'm learning something new for the first time, short-consistent intervals seem to make more sense. The time period would be anything within single practice session or out to 2-weeks of daily practice.

But there's a point of diminishing returns and eventually a time comes when switching to larger/increasing intervals provides more benefit. Possibly because more time is needed for the brain to effectively 'Chunk' all the new information into a more sophisticated understanding. So we get that effect where you've been struggling on a plateau before your vacation to Hawaii, only to find that when you come back you're now comfortably playing at a higher level.

AND THEN ... at that a point when you're playing comfortable at a higher level of skill - it might be more effective to go back to short term, consistent practice (single session, daily practice) to further explore the new skill, before continuing to move up the ladder.

(I'm not proposing any solutions, just pointing out the challenge of it all)

But I would like to propose a hypothesis:

1. When learning something new or fact based, it makes more sense to start with short-consistent intervals (within a single practice session or daily practice).

2. When chunking info into higher levels or understanding (like skills or procedures), it's more effective to use larger/increasing intervals (over several hours, days, weeks)

For some of us, learning from the score is much easier than learning something by ear.


One of the things I want to find out is if I experience playing music differently if I have it securely memorized. I also want to have some music that I can play if I find a piano somewhere, and don’t have music scores with me.


Yes, that’s what this general current interest on PW in spaced repetition is about.


I see the point of what you’re saying. For this experiment, I’m only ready to try it the rigid pre-scheduled way. Some things I’m after are consistency and spacing.

When I’ve tried to memorize things before, I’ve desperately repeated a phrase day after day, every day, and never feel confident. Then I miss a day or days, and everything falls apart because I never actually got the phrase out of short term memory so I never actually strengthened the memory by practicing recall of it.

Using the SRS gives me confidence that I won’t fall into those traps. Also it gives me confidence if I can’t remember a phrase, I just put it back to the beginning of Deck 1 and start the sequence of intervals over again for that phrase and — and this is key — that eventually I will indeed have the phrase learned and in long term memory. (Rather than my usual previous reaction, which would have been “oh I forgot this phrase, I’ve failed and I’ll never be able to succeed.”)

I’m thinking about the balance different students might need between flexibility vs. structure. For someone else, my precise rigidly scheduled approach on this SRS project might be too rigid (or too exasperatingly mechanical, or have too many logistics, or whatever problem). But for me right now it is a really useful structure that helps me have some important experiences that I personally don’t have the discipline or knowledge to access without the formal SRS.

I like what you’re saying overall, because I do find myself thinking about what kinds of tasks or learning would benefit from what kind of approach. I just don’t have any experience to be able to answer that for myself yet about most aspects of my practice.

Curious, I would think it’s exactly the opposite: facts take increasing intervals, procedures take every day.

It is even faster to look at the score in front of you, and get the music that way (there's more than a tune to piano music). You mention memory right after that. Why can't you memorize the tune right away from the score if you wanted to do that.

I did not manage to relate your post to the purpose of this thread: either "spaced repetition system", or the "other" Minuet in G. I understood each sentence, but did not catch the message.

Perhaps a better hypothesis would be:

In the beginning, rapid-short-consistent intervals of practice are more effective for learning, but reach a point of diminishing returns. Over the long term, larger/increasing practice intervals provide another set of benefits.

Both data and procedures benefit from better recall, but procedures have the additional benefit where the “chunking” of information seems to also get “better”.


... and I wonder if the different goals at the beginning, middle and end of learning a piece can help to select the appropriate repetitions/intervals.

Very well could be, Groove On. I feel like I’m incompetent at determining the spacing of repetitions on my own. I’m hoping to learn how to do this eventually, but for now I’m just working on staying consistent. E.g. with my flute longtones and piano sight reading, every day. With this Minuet, a predetermined schedule of increasing time gaps. And I’m working on another project, to learn musical intervals via SRS, that runs in a 28-day cycle.

With all of these, because they’re running on a predetermined system/schedule, I don’t have to think “am I good enough yet” or “should I take a break” or even “will I ever learn this... I don’t think I will... I’m a failure.” And I trust that the schedul s are good ones, that encourage consistency that will lead to progress.

I also haven't figured a practical way to calculate intervals. It's not clear to me how to apply them effectively. So I still just schedule things on a case-by-case basis.

The only effective 'system' I have is to cycle through my studies. I work on classical pieces until the brain starts to resist the monotony, then I switch to technical exercises, later my blues/jazz studies - and back around again. By constantly throwing changes at the brain it seems to keep me in a more mindful and active state, additionally, it naturally contains the SR effect.