Piano World Home Page Forums Our Most Popular Forums Piano Teachers Forum Piano practice with 'spaced repetition' systems (Anki)

As an 'ABFer', I wonder whether any folks on the Teachers forum have looked into using "Spaced Repetition" study systems as a part of piano practice? Instead of a straight flat schedule of rotation through, for example, scales and arpeggios etc etc, if a task is successfully completed, then the interval for the next review increases. I've been using Anki (it's free) for technical studies along these lines for over 6 months, with I believe, some positive results. However, I wonder if anyone else has gone down a similar route?

That website doesn't really explain this system of memorization. Can you please explain it a bit further for those of us who have nto heard of this method?

Hi Morodiene,

Certainly...

In general with a 'Spaced Repetition' system, you review studied material "just before you forget it". So for a new single study task, you may initially study it today for the first time (time zero), then again in 1 minute then in 10 minutes, then tomorrow and at subsequently increasing time intervals e.g 2, 5, 12 days. Clearly with increasing time intervals (in days) you may "fail" a task, in which case you start over.

In Anki, each study task has 4 possible answers, an out-and-out fail = "Again" (study it again now), then 3 degrees of pass "OK", "Good" and "Easy". It also uses these answers to calculate the "ease" of an individual study task. This "ease" moderates the interval calculation. So a task that is hard will have (even if answered "Easy") shorter intervals assigned than a task that is easy (no lapses and has always been answered "Easy").

So focus and study resource is given to what you can't do, whilst fluency and learning of what you can is maintained and grown (theoretically).

It sounds as if this practising is focusing on the angle of remembering or memorizing music or musical tasks. I'm focusing above all on physical skills, because all my hangup are in how I moved and held myself at the piano while self taught. For this you would not want to do this, I think. (or?) But if you wanted to remember scales and arpeggios in given keys, then a system helping memory might do it. So a) what goals are you looking for in your practising, b) what types of results are these positive results, more specifically? (remembering the scales, playing with greater ease because the time off has allowed it to come into your body, etc.)

<<It sounds as if this practising is focusing on the angle of remembering or memorizing music or musical tasks.>>

It's focusing transfering learning from "short-term" to "long-term memory" and increasing the length of time it's retained in Long-term memory. It makes you maintain these skills and knowledge through effectively practicing "recall" (at the increasing intervals). For me this practically feels to be "deep learning" or gaining "fluency" (and these sorts of apps are often used for language learning). Clearly if playing without a score it will also be "memorisation", but I have been using it whilst studying from a score and things I would never consider "memorising".

<<I'm focusing above all on physical skills, because all my hangup are in how I moved and held myself at the piano while self taught. For this you would not want to do this, I think. (or?)>>

It could well be "(or?)". I've been working on improving my weaker left hand and technique in general. For example: I had been cycling through the circle of fifths doing a detailed pattern of study on each scale: quarter, eigth, sixteenth notes; both HS & HT; in various rhythms and with time increasing the metronome value. This was taking a large chunk of my limited practice time. But I was bothered that it didn't feel "fluent" and when the metronome value got >90 it certainly wasn't without tension. With this emphasis on practicing recalling of learnt work, I have noticed an increasing sence of "fluency" and reduced tension, well that's how it feels subjectively.

<<So a) what goals are you looking for in your practising>>
Improve technique
Improve dexterity
Reduce tension
Play sixteenth note runs with greater ease particularly in the left hand.
Learn scales, arpeggios and chords around circle of fifths

<< b) what types of results are these positive results, more specifically? (remembering the scales, playing with greater ease because the time off has allowed it to come into your body, etc.)>>

1) Efficiency: The time spent on a given set of studies has reduced. So, for example instead of doing 5 minutes per scale and 15 minutes for a set of major and minors, I now do two minutes or so each scale sixteenths at current tempo, but for the next 30 days I've only got 12 days of this set of scales to practice. So I've been adding in additional technique study tasks. Some of these are not in the SR system, but will be!

2) Fluency: I feel I 'know' the study tasks better than following a simple rotation system.

3) Scope: Due to (1) I'm able to add more study tasks into the practice session and cover more ground.

Another concept to mention is "Maturity". Study tasks whose intervals are less than 21 days are considered "young" and still being learnt. So taking the scales example, currently 33 of 36 are "mature". This seems to be a bit of a watershed.

As an example here's the review history for my hardest and easiest scales 3-octave, sixteenth note HT, scale tasks at mm=72 (I dropt it down to something much more realistic for me).

Gmajor

Date Type Rating IntervalEase Time
2012-12-21 @ 12:37 Review 3 10.4mo 260% 1m
2012-08-23 @ 14:32 Review 4 4.0mo 260% 1m
2012-07-18 @ 15:13 Review 3 1.2mo 250% 1m
2012-07-04 @ 14:39 Review 3 14d 250% 1m
2012-06-29 @ 10:31 Learn 3 4d 250% 57.

G# minor melodic

Date Type Rating IntervalEase Time
2012-12-29 @ 17:34 Review 4 25d 155% 2m29s
2012-12-14 @ 18:17 Review 2 14d 140% 1m
2012-12-01 @ 17:36 Review 2 11d 155% 1m
2012-11-21 @ 20:26 Review 2 10d 170% 1m
2012-11-12 @ 14:46 Review 2 8d 185% 1m
2012-11-03 @ 13:18 Review 2 5d 185% 1m
2012-10-29 @ 19:45 Review 2 4d 185% 1m
2012-10-25 @ 17:59 Review 2 3d 185% 1m
2012-10-21 @ 17:01 Review 2 2d 185% 1m
2012-10-20 @ 14:50 Relearn 2 1d 185% 1m
2012-10-20 @ 14:46 Review 1 0d 185% 1m
2012-10-19 @ 15:26 Relearn 2 1d 185% 1m
2012-10-19 @ 15:23 Review 1 0d 185% 1m
2012-10-15 @ 19:50 Review 2 4d 185% 1m
2012-10-11 @ 19:37 Review 2 3d 185% 1m
2012-10-08 @ 14:34 Review 2 2d 185% 1m
2012-10-07 @ 12:54 Relearn 2 1d 185% 1m
2012-10-07 @ 12:50 Relearn 1 0d 185% 1m
2012-10-07 @ 12:46 Review 1 0d 185% 1m
2012-09-21 @ 22:01 Review 2 15d 205% 1m
2012-09-08 @ 18:11 Review 2 12d 220% 1m
2012-08-26 @ 09:32 Review 2 9d 235% 1m
2012-08-19 @ 10:11 Review 2 7d 250% 44.4s
2012-08-13 @ 17:40 Review 2 6d 250% 1m
2012-08-08 @ 18:26 Review 2 5d 250% 1m
2012-08-04 @ 09:07 Review 2 3d 250% 1m
2012-07-31 @ 17:18 Review 2 2d 250% 1m
2012-07-29 @ 08:50 Review 2 2d 250% 1m
2012-07-28 @ 08:32 Relearn 2 1d 250% 1m
2012-07-28 @ 08:30 Review 1 0d 250% 1m
2012-07-27 @ 17:16 Relearn 2 1d 250% 1m
2012-07-27 @ 17:13 Review 1 0d 250% 38.4s
2012-07-26 @ 16:47 Relearn 2 1d 250% 1m
2012-07-26 @ 16:45 Review 1 0d 250% 29.0s
2012-07-23 @ 15:28 Review 2 2d 250% 1m
2012-07-21 @ 10:10 Review 2 2d 250% 1m
2012-07-20 @ 17:53 Relearn 2 1d 250% 1m
2012-07-20 @ 17:43 Review 1 0d 250% 56.3s
2012-07-15 @ 08:52 Review 2 5d 250% 1m
2012-07-11 @ 17:39 Review 2 4d 250% 1m
2012-07-08 @ 08:59 Review 2 3d 250% 1m
2012-07-05 @ 19:50 Review 2 2d 250% 1m
2012-07-03 @ 18:04 Review 2 2d 250% 1m
2012-07-02 @ 17:39 Relearn 2 1d 250% 1m
2012-07-02 @ 17:24 Review 1 0d 250% 1m
2012-06-30 @ 08:21 Review 2 2d 250% 36.8s
2012-06-29 @ 10:46 Learn 2 1d 250% 1m
2012-06-29 @ 10:43 Learn 1 0d 250% 1m

This one is interesting, 'cos it seemed to go through
several cycles of failure, before 'ignition' and taking
off.

Please note though that "1m" was the maximum time in Version 1.x of Anki and why the more recent times are 2m or so.

For the next 30 days, I have 23 scales to practice which
assuming no failures works out at <60 minutes.

The current goal is to get all the scales to status mature
(intervals >=21 days)....

But it's not just for scales or similar technique studies...

I'd rather spend time practicing than rating and keeping track of all that stuff.

My system goes like this:

A. Play that bit again
which results in
1.Yah! that was better, do it one more time.
or
2.Oh, that wasn't better, do it again.

There is no doubt that the spaced recall systems are very effective for remembering information (academic courses, etc.)

I haven't heard of people using it for physical skills so this is interesting.

I think the problem for many people may be the amount of structure required. Some of us are comfortable with that, and maybe better suited to that type of programmed system. I suspect more of us are not, and would rebel even if it is proven to work.

Hi Malkin,

<<I'd rather spend time practicing than rating and keeping track of all that stuff.>>

I agree, however the data is gathered "under the hood". It's very unintrusive and non-invasive in practice.

Hi TimR,

<<I haven't heard of people using it for physical skills so this is interesting. >>

It was this aspect that interested me. In biological terms, are not many "physical skills" really neuronal based? There is a lot of disinformation e.g "Muscle memory" muscle's don't have memory, only nerves have memory. Whilst I'll admit that physical training leads to muscle development and muscle fibre type modification depending on the training mode. My personal opinion is that many of the patterns we practice and learn are largely "Neuronal" with both peripheral nervous system and central nervous system components. So my thoughts were that it may be a misnomer to consider "physical" and "mental" skills learning as separate systems. If so then study techniques for mental skills may be equally applicable to physical skills.

Do you mean that after I download Anki onto my computer or tablet Anki will keep track of my practice and progress?

I don't have to do anything except practice?

Hi Malkin,

<<Do you mean that after I download Anki onto my computer or tablet Anki will keep track of my practice and progress?

I don't have to do anything except practice?>>

Yes, download, then install. Then set up a set of study tasks.
I hear what you are saying.

To set up a set of study tasks you can either browse sets that others have created and download and use those, or create your own. Initially I downloaded some music related 'decks' to get the feel of how it works.

There a number of ways of entering new tasks, one task at a time or to import from a file. For small 'at piano' study sets for scales, I used Notepad to create a list, then imported it.

The process is to select a set of study tasks, click "study", then work through each task in turn. When each task is complete you click "Show answer", for which I use "OK or not?". As I mentioned above, you have a choice of 4 answers: Again, hard, good and Easy. It then moves onto the next task, when the last is complete it returns to the list of study task sets ('decks' in Anki).

I accept that there's time involved in becoming familiar with the software and the process. However, at the piano it is really quite quick and painless. In Anki you can 'sync' study sets across multiple PCs and platforms, you can also access your personal study sets via the internet.

So at the piano, I use an Android tablet on the music stand and access my study sets via Ankiweb, you could also use a Smartphone, iPad etc etc. I manage and review the study sets away from the piano at the PC and at times when I'm not able to get to the piano.

I fully accept what TimR points out that some may not accept the structured approach or the use of software during practice (and using Anki is only a part of my practice routine).

EJR, I'm not opposed to something like this if it actually works for music. Certainly it sounds like it works well for memorization, which can be very helpful in and of itself. However, I'm not really convinced that it works for other tasks that don't rely upon memorization.

Scales, for example, have long since been memorized for me, and if I were trying to increase speed or accuracy, how would one determine the task that is to be repeated in these intervals? If I'm playing them at 120 bpm and I'd like to get them up to 144 bpm, for example, if I jump in and start playing at 144, I will most likely fail, as there is a process to increasing the speed that I've skipped over, and no matter how much I try to do it faster it won't be good if all I am used to doing is 120. On the other hand, if I do them at 120, then I'm not improving since I can presumably already do them at that speed. Lastly, if I do this series of intervals over 12 days starting at 120, then increasing to 126 for 12 days, then 132 for 12, etc., we are looking at very long time to increase speed when compared to the standard way of doing this (a couple of months vs. a couple of weeks the standard way).

This is just one example, but it applies to many technical tasks. I am someone who really looks for efficiency in practice, so I'm not a proponent of blind repetition by any means. I just am not convinced this particular method can be applicable to this particular kind of task based on what I know of traditional methods.

Hi Morodiene,

<<scales for example>>

I wish I could play them at 120 bpm! You raised a number of interesting issues.


a) Consolidation and Maintenance.
"Playing them at 120". If this is your current standard, then if you fed it through Anki and say defined a set of study tasks:

"Play Scale X at mm=120 over 4 octaves HT 5 repetitions".

If you can currently play every scale then the answer for each would be "Easy". The new interval for the next review would basically be 2.5x the current interval. So it would go something like:

1 day, 2.5 days, 7 days, 15 days, 1 month, 2.5 months, 7 months, 1.6 years, 3.5 years etc etc or thereabouts. I suppose this is similar to my G major example above.

"2012-12-21 @ 12:37 Review 3 10.4 mo"

So I last reviewed it on 21/12 and the next looks to be mid-October this year ;-)

b) Improving a standard (e.g. tempo).
I think the target standard for each task has to be carefully selected for the individual and achievable so if you are currently playing at 120 bpm then perhaps something like:

"Play Scale X at mm=120 to 132 step 6 over 4 octaves HT 5 repetitions at each"."

There may be some scales that go well at the higher tempos, some that are doable but hard and a number that are outright fails.

In this situation you are combining both maintenance (of those that were OK at the new target of 132) with learning and focussed work on those that failed at either 126 or 132.

When all are 'mature' you would increment the attribute of the study set to say mm=136 step 4 (136, 140 and 144).

Yes, but my point is, I could do that in a matter of weeks with how I practice now, vs. months that this system seems to promote. To me, that's not as effective.

Hi Morodiene,

<<I could do that in a matter of weeks with how I practice now, vs. months that this system seems to promote. To me, that's not as effective.>>

I agree that it's not ideal for short term gains. But what about efficient maintenance of acquired skills?

Back to the OP " I wonder whether any folks on the Teachers forum have looked into using "Spaced Repetition" study systems as a part of piano practice?"

I guess it's a "No" then...

No, but it doesn't mean that we aren't interested in learning about it. Don't mistake my questions as dismissal of the method altogether. So really this method is not intended for learning new things, but more for retention.

Ideally, this is something that could be applied to maintaining repertoire that has been memorized over a long period of time. At least that is what I'm getting from it. I'd be curious to try it out in such circumstances and see what happens.

Hi Morodiene,

"not intended for learning new things"

I have tried it for easier pieces I was trying to learn in a target of a week (or so) and as you suggest it offers little benefit in this situation.

However, if the time frame for learning a piece (or other skill) is longer (say several months minimum) then I think it may well be of use.

The closest example that I've presently got to this is as a deck of "Micro-studies" derived from Chopin Etude 10/4. Currently I have no plans to 'play' it end-to-end, only study it! There are 30+ cards, with around 10 left hand, right hand and short HT phrases or sections. It's easy to incorporate images (and audio) into the "question" in the study note. Once the deck is "synced" then it's available on all my devices. At the piano I use the tablet in landscape mode, and in this example has replaced the score. The latest version of Anki features improved " Learning modes", so you can cycle each study tasks as many times as necessary. I think this is an example of learning something "new" - See screen shot below.


"is something that could be applied to maintaining repertoire that has been memorized over a long period of time"

You raise a very good point regarding maintenance of repertoire. It would be ideal for this. For easier pieces you could simply add them to a "My repertoire" collection of tasks, as they are learnt. Anki would then present them for review at increasing time periods. This is something that is definitely on my list of things to do.

For maintaining learnt "challenging" pieces, each having it's own deck, then you could also make an entry for these in the "My repertoire" set. However, appropriate whole piece performance tasks can be included in the deck for the piece itself. You could always do both (keep the challenging piece study deck going and add an entry to 'My Repertoire' set).


An example of something "New" - Chopin 10/4 "Micro-studies"



This example has had several review cycles of study (the 7, 15 & 25 days indicate this). With every review, these descriptors change. I'm NOT trying to memorise these phrases.

I'm only aiming to play these slowly at present (note learning). I was looking for something to give my left hand a work out and particularly the weaker fingers and this seems to fit the bill perfectly. This seems to have gone OK, so I added in those those tasks for the right hand and HT. Currently all the "HT" tasks are "suspended" and not yet being studied.

This is my second attempt at this 'deck'. The first was just 'text' as in the description for scales previously in the thread. But the latest version of Anki has improved handling of media files. This means that media in tasks are immediately available via the Ankiweb interface. Preparing the images took half an hour or so, but I see this as useful pump-priming.

If folks were to incorporate images of the score, then it would probably be best to consider a 10 inch or so Android tablets or iPads for "at the piano" studies.

Decks can also be publicly shared or exported so there's the possibility of collaborative projects/social aspects to all this.

Thanks for sharing this, EJR. I find it very interesting and I'd like to hear more about your progress with it. Would you be willing to let us know how you progress with this etude?

Sure, I'm very happy to report back on progress on these "Micro studies" derived from the Etude ;-) (And who knows depending on how these progress? ...well maybe one day).

Well, I do tell my students to essentially do microstudies of a different kind, but you put enough micros together and you have a whole piece .

I use SRS extensively for second language reviewing, and I've also Anki for music theory/history studying in the past.

One thing to remember with any SRS system is that it presents each card at (hopefully) just the right point before you forget it again. The net effect is that you're barely remembering a lot of things. It's a handy memory tool for sure, but to master the material, you'll need to use your material more extensively in other contexts.

That being said, I believe an SRS system would be a great way to review repertoire and technique that you have already learned well (though playing it more often than minimum would be even better).