Swing Playback - Notes affected by

[Doug Kerr]

This is to update my discussion of what notes will have their start or end play time adjusted by the Swing Playback function in Encore.

Definitions

In what follows, "note" means:
• A simple note
• An augmented notes (i.e., one with a dot or double dot)
• A group of notes tied so they behave as a longer note

"Basis note" means the note value that is set in the Swing Playback dialog.

The basic rules

A. If a note:
• Has a face value as long as the basis note or longer, and
• The end of its face value falls on an even-numbered count of a series of basis notes *

then the Swing function will delay its play end time.

This could be the first note of a pair of notes to be "swung".

B. If a note:
• Has a face value as long as the basis note or longer, and
• Falls on an even-numbered count of a series of basis notes *

then the Swing function will delay its play start time time.

This could be the second note of a pair of notes to be "swung".

* "Even numbered count of a series of basis notes" means, if the basis note were set to eighth note, if we imagine counting at an eighth note pace starting at the the first beat of the measure, the instants of the 2nd, 4th, 6th, etc. count.

Other ways to state these rules

Readers will recognize that one or both of these rules can be restated in ways (entirely equivalent) that may best match their vision of the application of this doctrine to musical practice . Here are ones often cited:

A1
a. If a note:
• Has a face value the same as the basis note , and
• Falls on an odd-numbered count** of a series of basis notes [** variation of definition above]

then the Swing function will delay its play end time.

b. If a note:
• Has a face value longer than the basis note, and
• If we consider it to consist of two parts tied, the last one of which has the face value of the basis note, that last part starts on an odd-numbered count of a series of basis notes *

then the Swing function will delay its play end time.

For the case where the basis note is set to eighth note, as is common in the application of swing:

A2. If a note:
• Has a face value of an eighth note or longer, and
• Falls on a beat of the measure ***

then the Swing function will delay its play end time.

B2. If a note:
• Has a face value of an eighth note or longer, and
• Falls on an "and" count of the measure (halfway between beats) ***

then the Swing function will delay its play start time.

*** "Beat" here means beats at the rate of a quarter note (as is commonly true in meters where the denominator of the time signature is "4").


These alternate statements of the rules are exactly equivalent to the "basic rules".

Doug

[polarbreeze]

This is to update my discussion of what notes will have their start or end play time adjusted by the Swing Playback function in Encore.

Definitions

In what follows, "note" means:
• A simple note
• An augmented notes (i.e., one with a dot or double dot)
• A group of notes tied so they behave as a longer note

I think that the definition of "note" needs to include "rest" because a note-rest pair or a rest-note pair must also be swung if they meet the other criteria.

Also, I think that what you call "augmented notes" are disqualified because the convention is that if the writer explicitly shows modified timing (ie anything other than strictly written equal-value pairs), his/her intention is to overrule the general swing instruction. Note that the swing modifier must only be applied to a pair of notes (or note/rest pair) in which both notes are written with the same value.

With respect to the eligibility of a "group of notes", in general I think groupings are excluded for the same reason as above (ie the writer's specific instructions prevail). But a note that is otherwise eligible as part of an equal-value pair is not disqualified just because it happens to be tied to another note before or after. The swing adjustment is then applied to that specific note's face value, not to the value of the tied group.

By the way, I have to say that I think it's very difficult, and perhaps artificial, to try to come up with generalized rules/definitions for this. I don't think I've ever come across a swing situation applied to anything other than eighth-note pairs so I find it hard to avoid going back to that specific.

[Doug Kerr]

Hi, p.

This is to update my discussion of what notes will have their start or end play time adjusted by the Swing Playback function in Encore.

Definitions

In what follows, "note" means:
• A simple note
• An augmented notes (i.e., one with a dot or double dot)
• A group of notes tied so they behave as a longer note

I think that the definition of "note" needs to include "rest" because a note-rest pair or a rest-note pair must also be swung if they meet the other criteria.

A good reminder, but . . .

Remember (despite our inclination to speak of it that way) the algorithm not work on pairs - just on notes, individually, and probably does not directly work on rests.

If we have a "rest-note pair", or a "note-rest pair", the note is affected per the rule. And thus the actual duration of the rest adjacent to the shifted end changes. But not because the rest is "adjusted" by the algorithm.

Perhaps the best thing would be some narrative reminding the reader of the impact of this on rests.

I'll certainly mull that over.

Also, I think that what you call "augmented notes" are disqualified because the convention is that if the writer explicitly shows modified timing (ie anything other than strictly written equal-value pairs), his/her intention is to overrule the general swing instruction. Note that the swing modifier must only be applied to a pair of notes (or note/rest pair) in which both notes are written with the same value.

Note that this was not a discussion of notation practice. It was a presentation of the rules followed by Encore. And its musical education is so limited that it does not honor the principle in red. So while you, and music teachers, may disqualify them, Encore does not.

To give an example not involving augmentation, if we have an eighth note followed by a quarter note, and the eighth note meets rule A, it will, be adjusted, and if the quarter note meets rule B, it will be adjusted.

With respect to the eligibility of a "group of notes", in general I think groupings are excluded for the same reason as above (ie the writer's specific instructions prevail). But a note that is otherwise eligible as part of an equal-value pair is not disqualified just because it happens to be tied to another note before or after. The swing adjustment is then applied to that specific note's face value, not to the value of the tied group.

Your description in terms of a musical vision of the process is very apt. (One of my "alternate rules" was an attempt to articulate that vision.)

But in fact the rules as stated (as will be the ones I will shortly state on the amount of time adjustment) are accurate without resort to visualizing some part of the "tie" being isolated for treatment (even though that model can be equally valid). Now it might work that way inside Encore (and I'm looking into that).

By the way, I have to say that I think it's very difficult, and perhaps artificial, to try to come up with generalized rules/definitions for this. I don't think I've ever come across a swing situation applied to anything other than eighth-note pairs so I find it hard to avoid going back to that specific.

That sounds really good. Note that this would be a wholly different exercise from what I did in this article, which was to report on what Encore does (for better or worse).

Keep in mind that I do not do this as a substitute for a "musical" outlook on the process. Ultimately the latter is what is important to the user, not the propeller-head.

Unless we actually know what the rules follows by Encore are (which has to be done in a rather dry technical description), we cannot hope to properly recast those rules in a "meaningful to the musician" basis. For example, you well stated some rules imposed by musical teaching. But Encore doesn't honor them. If we don't make the "practical" expression of what Encore does comport with the "technical" description, the user may get a bad surprise. So we must have the uncolored technical description in hand first.

Thanks again for your insight.

Doug

[Doug Kerr]

Hi, q,

Well, although I was somewhat "dismissive" of (while in the "dry scientific context) the view of parting off the part of a long subject note equal to the basis note, I realize that when it comes to actual times, taking duration percentages into effect, that is exactly what effectively happens.

That of course does not make the rules as I stated them in they article under discussion incorrect - from a 'black box" perspective they still are.

By the way, my proposed "dry" statements as to how start and end times are changed do not depend on that "vision" either for their context, but in fact it can be seen in the mathematics!

So I need to give some further thought to my manner of presentation.

I also find that my "proposed" algorithm does in fact not match Encore's work (putting aside the anomalous area altogether) and I need to reconsider it.

I'm doing some more testing to make sure I have all the pertinent facts.

Thanks again for your insights.

Doug

[Denkster]

This morning, in pursue of a valid model of our relevant reality*) - in casu quo musical practice of Swing Playback - I consulted a percussion teacher / big band drummer.
He emphasized the point that 'swing feel' means:

the one that starts close to the pulse - you call it the second note - is delayed
For drummers, the playback duration of notes is no issue.
He writes swing eighths like this: a tuplet of (1/8 note + 1/8 rest +1/8 note).

He lent me this book:

John Laporta; "Developing Sight Reading Skils in the Jazz Idiom"; Berklee Press Publications 1967.

Unfortunately it is out of printing and I can't find it on the second hand market either.
My friend the percussion teacher had to use this book in his study at the Royal Conservatory in The Hague, decades ago.
The book is full of songs with phrasing suggestions.
Here are some.

swing_playback-examples-1.jpg (59.54 KiB) Viewed 2109 times

swing_playback-examples-4.jpg (150.92 KiB) Viewed 2103 times

swing_playback-examples-8.jpg (147.63 KiB) Viewed 2100 times

Another interesting thing: 'swing sixteenth' is done. He showed me a course in Caribbean style big band drumming, with a song that started in a time signature of 9/8, to be played with swing sixteenths, and halfway switched to 12/8, with swing.
(Edited later) I forgot to mention: sixteenth notes that swing in a 4/4 measure are common in funk.

*) A model of our relevant reality must be at the base of every design. IMHO, 'Relevant' is the most difficult part of this statement, because that commands us to determine the scope of our project.

Kind regards
Eveline

[Doug Kerr]

Hi, p,

In fact, this vision:

For an eligible note longer than the basis note, the adjustment can be thought of as affecting the portion (at the affected end) that is the basis note in length.

turns out to work for all aspects of the adjustment.

And as you noted, it works out for "eligibility" as well.

Based on that, I think I will consider stating the rules regarding time details in that context.

There is also a way to restate the rule for eligibility. It could be done this way (as a first cut):

A. If a note:
• Has a face value as long as the basis note, or longer, and
• If the start of the last portion of the note, equal in length to the basis note in length (for a note exactly the length of the basis note, this is just the start of the note), falls on an odd count of basis notes.

the end time of the note is delayed.

(I think the parenthetical is needed to avoid befuddling the reader as to the meaning of "the last portion of the note . . ." if the entire note were exactly the basis note in length. Again, it would not be in any way ambiguous, but it might throw the reader off his pace.)

This is to me more indirect, and a bit more cumbersome, but if it is helpful to interpretation, I'm up for it.

There would seem to be no reason to fool with Rule B; the equivalent would just seem silly. It would speak of the start of the first portion of the note, rather than of the start of the note.

Note that, so far as Encore's behavior is concerned, it matters not how the note is composed.

If, for example for some bizarre reason we were to compose a "dotted quarter note" with twenty-four 32th notes tied, then this would still qualify (for a basis note of eight note) if (and only if) the "face end" of the last 32nd note fell on an even-numbered 16th note count; the start of the last 32 note isn't of interest. (We know this from my original statement of the rule, which is still accurate.)

My swing algorithm

I had said that my suggested "reformed" swing algorithm comported with Encore's behavior with respect to "first notes". In fact that is not true (except for notes corresponding to the basis note).

My algorithm could be said to "preserve the duration fraction" of the first note when it was extended.

In fact the Encore algorithm could be said to "preserve the duration fraction of the last portion of the first note, having a length equal to the basis note" when the first note was extended.

There is no simple criterion by which we might decide which one of these (if either) is "best".

For the moment, I will recast my suggested algorithm to comport with Encore's behavior with regard to "first notes", and will make a corresponding modification to the doctrine for "second notes".

Again, of course, this is further "vindication" of your enthusiasm for the "last chunk" outlook. I will certainly make sure it is kept in view, where appropriate.

Thanks again for your insights.

Doug

[q]

Hi, q,

Well, although ...

Doug

Hi Doug,

Was this message directed to me? Or are you getting your ps and qs mixed up?

q

[q]

Another interesting thing: 'swing sixteenth' is done. ... in Caribbean style big band drumming ... [/list]

Also, there is a large collection of Brazilian Choro music (in fact one of Eveline's "online Encore score links" has lots of Choro music.)

Choros are usually written (not sure how to express this) with sixteens where we'd write eights. And I believe a considerable amount of the repertoire is to be interpreted with swing. I know a couple of experts in the field, so I could ask.

John

[Doug Kerr]

As a result of some useful discussions with our colleague "polarbreeze" on this topic, I have revised this "report", as follows:

************

This is to update my discussion of what notes will have their start or end play time adjusted by the Swing Playback function in Encore.

Definitions

In what follows, "note" means:
• A simple note
• An augmented note (i.e., one with a dot or double dot)
• A group of notes tied so they behave as a longer note

"Basis note" means the note value that is set in the Swing Playback dialog.

The basic rules

A. If a note:
• Has a face value as long as the basis note or longer, and
• The end of its face value falls on an even-numbered count of a series of basis notes *

then the Swing function will delay its play end time.

This could be the first note of a pair of notes to be "swung".

B. If a note:
• Has a face value as long as the basis note or longer, and
• Falls on an even-numbered count of a series of basis notes *

then the Swing function will delay its play start time.

This could be the second note of a pair of notes to be "swung".

* "Even numbered count of a series of basis notes" means, if the basis note were set to eighth note, if we imagine counting at an eighth note pace starting at the the first beat of the measure, the instants of the 2nd, 4th, 6th, etc. count.

Other ways to state these rules

Rule A1 can be equally-accurately stated in another form that more directly relates to the way we usually think about swing in practice. I will do it in two pieces since the "universal" description can be a bit confusing.

A2a. If a note:
• Has a face value the same as the basis note , and
• Falls on an odd-numbered count ** of a series of basis notes [** concept defined above]

then the Swing function will delay its play end time.

A2b. If a note:
• Has a face value longer than the basis note, and
• If we consider it as if it consisted of two or more notes tied, the last one of which has the face value of the basis note, and that last part starts on an odd-numbered count of a series of basis notes

then the Swing function will delay its play end time.

Going even further in the direction of practical musical application, for the specific case where the notes of interest are eighth notes, and the basis note is set to eighth note, as is common in the application of swing, the rules become:

A3. If a note:
• Falls on a beat of the measure ***

then the Swing function will delay its play end time.

This could be the first note of a pair of eighth notes to be "swung".

B3. If a note:
• Falls on an "and" count of the measure (halfway between beats) ***

then the Swing function will delay its play start time.

This could be the second note of a pair of eighth notes to be "swung".

*** "Beat" here means beats at the rate of a quarter note (as is commonly true in meters where the denominator of the time signature is "4").

All these alternate statements of the rules are exactly equivalent to the "basic rules".

Rests

If there is a rest adjacent to the "business end" of an eligible note, we do not need to think of it as being affected by these rules, as such (in fact, technically, we really cannot).

The impact on the rest is just that which unavoidably occurs from the change in the time of the adjacent end of the adjacent note.

And in fact, whether a rest is affected cannot be determined from the properties of the rest. (There is, for example, no minimum length of rest that may be affected by swing.) The effect on the rest is only a byproduct of the affect (if any) on the adjacent note, and its eligibility for that is determined on the basis of the note's properties alone, as described above.

Of course, if we have an eighth rest followed by an eighth note eligible under Rule B, and apply swing, the eighth rest is prolonged. And that may well be an important vision from a musical perspective. But it does not require an additional rule, or even an enlargement of the terminology of the existing rules, to accommodate that.

If we feel the need to state rules for rests, here they are:

X. If for a rest:
• The end of its face value falls on an even-numbered count of a series of basis notes

then the Swing function will delay[/b ]its [b]end time.

That means of course only that the start time of the following note has been delayed, as discussed above.

Y. If for a rest:
• It falls on an even-numbered count of a series of basis notes

then the Swing function will delay[/b ]its "[b]start time".

That means of course only that the face end time of the preceding note has been delayed, as discussed above.

In the next chapter of this report, where I will speak about the actual amount of time changes, we will find that trying to discuss the effect on rests of the swing process would become truly bizarre. (And remember, no MIDI messages are sent for rests, as such. There is no Rest OFF message, and we will be badly misled if we think about when it should be sent. And a Note ON message is not ncessarily equivalent to "Rest OFF".)

So, while I well understand the musician's outlook that, if we have an eighth rest followed by an eighth note, both are "affected" by the application of swing, when stating technical rules (as we propeller heads try to do at the "top level" of these matters), it is best to realize that what happens to rests is an unavoidable result of what happens to notes, and only state rules for the notes.

Doug

[Doug Kerr]

Hi, everybody,

Hi, q,

Well, although ...

Doug

Hi Doug,

Was this message directed to me? Or are you getting your ps and qs mixed up?

Probably for "p".

We are all exhorted to mind ours, but I don't always!

Sorry.

And thank you so much for not saying "p's and q's".

Doug

[polarbreeze]

In what follows, "note" means:
• A simple note
• An augmented note (i.e., one with a dot or double dot)
• A group of notes tied so they behave as a longer note

Sorry but I still object to this definition for the same reasons as before. For example, dotted notes are not to be swung because the writer's intention is to have them played as-written. Only note pairs or note/rest pairs of equal written duration are to be swung.

A. If a note:
• Has a face value as long as the basis note or longer...

I guess that's what the manual says and maybe that's the way it actually works now but I don't think this is how "swing" is interpreted in real life (which hopefully is what Encore is trying to emulate). I think the "or longer" is wrong.

[q]

Was this message directed to me? Or are you getting your ps and qs mixed up?

Probably for "p".

... thank you so much for not saying "p's and q's".

You are very VERY welcome.

Excellent contributions of recent. Many thanks to you and polarbreeze for hashing out swing things in my absense; I've been pretty sidetracked recently.

One thing I keep wondering though ...

Matt, although the recent 'swing' discussions have been many, far flung, winding, and sometimes backtracking, are they of any value to you?

In the future, when an important or lengthy discussion ensues, would you step forth to steer the direction somewhat, by expressing interest, by belaying, or by revealing critical concept to which only you may be privy? I' believe that our energy can be more effectively employed with direction and opinions from you, and Hotch.

Any interest in the sandbox forum idea? ... and area where we can hash things out before posing a condensed report in the beta or request areas. That would allow you to more easily focus on our best efforts without having to read every message, and yet the threads would be there (in the sandbox) in case you want to review the discussions that led to our conclusions.

A wiki's probably a better idea, if someone wants to set up one that's Mac and Windows friendly.

q

[Doug Kerr]

Hi, p,

In what follows, "note" means:
• A simple note
• An augmented note (i.e., one with a dot or double dot)
• A group of notes tied so they behave as a longer note

Sorry but I still object to this definition for the same reasons as before. For example, dotted notes are not to be swung because the writer's intention is to have them played as-written. Only note pairs or note/rest pairs of equal written duration are to be swung.

Please remember that what I have published is a description of how Encore works, not limited to situations consistent with good notation or performance practice.

If in fact, with the basis note set to eighth note, we have a group of notes tied, and:
• their effective face value is equal to or greater than an eighth note, and
• the end of the face value of the tie falls on an even-numbered "count" in eighth notes

then the end play time of this tie will be delayed by Encore. It really will.

This may make the music teacher in you want to crack Encore across the knuckles with a ruler, but it's a fact.

And recall it is up to the user to apply the swing adjustment to certain measures. We might hope that he would refrain from doing so where current pedagogy suggests otherwise, just as we might hope that he would avoid other things that today, in many circles, are consider musical faux pas.

But none of that changes what Encore, in today's implementation, actually does.

A. If a note:
• Has a face value as long as the basis note or longer...

I guess that's what the manual says and maybe that's the way it actually works now but I don't think this is how "swing" is interpreted in real life (which hopefully is what Encore is trying to emulate). I think the "or longer" is wrong.

Please remember that this is intended to be a description of how Encore works. I am not trying in this report to describe what we think might be a good thing for Encore to emulate (although that is an important project to which we should probably apply our abilities). I can't arbitrarily misstate what actually happens just to avoid mention of a situation in which common musical teaching indicates that the swing technique should not be applied.

It would be like not mentioning in a report of the measured response of an automobile speedometer what it would read for an actual driving speed of 120 m/hr because in must jurisdictions in the U.S. it is not legal to drive at that speed.

Doug

[Doug Kerr]

In my report on what notes were affected by the Encore Swing Playback function, I left the reader hanging with "then, the end time of the note is delayed" or such.

Now I will discuss "how much".

The Encore Swing Playback algorithm is extremely peculiar, especially in certain "zones". Here, I will only describe its behavior in those situations where the result can be traced to some rational "strategy". Fortunately, this embraces many of the cases of practical interest.

I will a little later describe the results on a "verbal" basis. But that is very complicated to do in a general way; it is best done with specific examples.

Before I do that, I'm afraid I must resort to a mathematical presentation description (again, limited to the regimes of "rational" behavior).

To do that, I need to define some symbols for quantities:

B - This is the face duration (in any handy metrical time unit) of the "basis note", the note we choose in the Swing Playback dialog.

S - This is the "swing parameter", the numerical value we set (as a percentage) in the Swing Playback dialog. It is in the unit percent, and can run from 55 to 75.

D - This is the "duration fraction" for the note prior to the application of swing: its sounding length described as a percentage of its "face length". It is in the unit percent. Common values are from perhaps 75 to 100. We assume that, initially, the start time of the note is at its "face start" (which will always be true for notes entered directly, not via a MIDI file or live recording).

THE MATHEMATICAL DESCRIPTION

1. "First notes"

By "first note", I mean a note that is eligible for the application of swing adjustment under "Rule A" in the first part of this report. In order to limit us to the "rational"parts of the algorithm, we will assume that all the notes in which we are intersted, and that become eligible for the application of swing adjustment, have the same face value as the basis note. (This is typical of actual operation.)

This is typically the first note of a pair of notes to receive swing adjustment (although a solo note meeting the eligibility criteria will be handled exactly the same way).

If the eligible note has a face value greater than the basis note, the rule is complicated, and some parts of it are not yet known to me.

Start time

The start time of the note is not affected.

The start time of the note is not affected

End time

For such a note, the application of Swing delays its actual end play time by this amount (in the same time units used for B):

B*D*((S-50)/50)

2. "Second notes"

By "second note", I mean a note that is eligible for the application of swing adjustment under "Rule B" in the first part of this report.

Start time

For such a note, the application of swing delays the start time by:

B*((S-50)/50)

End time

To keep to the "rational zone", we must only treat notes whose original play duration is 100% of the face value.

For such notes, the application of swing makes no change in the end time.

THE VERBAL DESCRIPTION

We will only describe the exact result for certain specific situations. Again we are speaking only of the affected notes being eighth notes, and the basis note is set to eight note.

First notes

Start time

The start time is unaffected

End time

Case 1: Initial duration fraction = 100%, swing parameter = 75% (the maximum possible)

The end time of the note is delayed by half the length of an eighth note.

Case 2: Initial duration fraction = 100%, swing parameter = 66.7%

The end time of the note is delayed by one third the length of an eighth note.

Case 3: Initial duration fraction = 900%, swing parameter = 75% (the maximum possible)

The end time of the note is delayed by 45% of the length of an eighth note.

Case 4: Initial duration fraction = 900%, swing parameter = 66.7%

The end time of the note is delayed by 30% of the length of an eighth note.

Second notes

Start times

Case 1: Swing parameter = 75% (the maximum possible)

The start time of the note is delayed by half the length of an eighth note.

Case 2: Swing parameter = 66.7%

The start time of the note is delayed by one third the length of an eighth note.

End times

Case 1: Initial duration fraction = 100%

The end time is unchanged.

Case 2: Initial duration fraction = less than 100%

The end time will in general be changed, sometimes delayed, usually made earlier. The exact rule is as yet unknown to me.

Doug

[Doug Kerr]

Hey, p,

Well, it turns out that in the Encore swing algorithm, "first notes" department, it seems like your concept of "working on the last basis note's worth of the note" in fact happens, with an interesting twist.

Let me call that "last basis note's worth of the original note's * face value" the "target cell".

* Sorry, but here "note" must include augmented notes and tied groups. If you find that intolerably offensive, stop reading.

Assume for now that the end time of the original note falls within the target cell. (It of course might not, depending on the original play duration fraction of the note.)

Now suppose we apply swing with a parameter of 75%. The result (looked at from the perspective of the "target cell") is to lengthen the target cell by half the length of the basis note. Thus, the length of the target cell is increased to 150% of its original size. We of course cannot see or hear this - it is a totally artificial thing, like a "construction line" in a geometric proof. But its role will show up shortly.

Now consider the amount of the actual original play time of the note that fell inside the target cell. How much that is depends on the original play fraction; if that fraction is too small, there is none inside. But were are not dealing now with that possibility.

In any case, after the swing, the note end time is set so the amount of play duration inside the target cell is 150% of what the amount that was inside the target cell initially.

Said another way, using the start of the target cell as our "anchor", the end of the target cell (something we cannot see or hear) and the actual end sounding time of the note are moved out proportionately.

I find this to be a very tidy vindication of your enthusiasm for the notion that swing works on what I here call the "target cell".

But now it gets a little funny. What about notes that were so short in original play duration fraction that there was none of their play duration in the target cell? The rule above can't be applied to determine their new end times.

Well, then evidently Encore shifts into another mode. The end times for such notes are delayed, but the pattern is not immediately apparent. I may be able to crack the function, or I may lose interest.

Perhaps this aspect of the algorithm is a "poison pill" to punish those who abuse the swing function by applying it to notes of forbidden length.

Very interesting.

Doug