Swing playback problems, start/end offsets and all

[Michael_Kilpatrick]

I’m going to describe what I would call a bug as I don’t think it’s the intended behaviour and is something I’m extremely disappointed with. I put a post on the Facebook group about this but here it is for the benefit of more of the Dorico team.

In my scores I will need a LOT of note duration alteration if I want to get a pleasant playback experience. It’s a fundamental aspect of my “urtext” Duke Ellington transcriptions. In a vast number of cases, off-beat swung notes whether written as a crotchet, two tied quavers, or a quaver tied to a crotchet, are all short notes in most contexts. Using Sibelius for many years this has been very easy to deal with - one can set the playback duration of each and every note without affecting other characteristics.

Dorico has that facility. It can be done either graphically in the Play “piano roll” or numerically using the Properties panel in Write mode. One can set a Playback End Offset and/or a Playback Start Offset. But it is broken.

If you drag the end position of an off-beat (swung) note in the piano roll, the start position of the note is instantly snapped back to the 50% (straight) position. If you do the operation using the Properties → Playback End Offset you don’t see the problem happen until you try playing it back because the Playback Start Offset remains disabled (unless you have touched it).

After switching off in a huff last night, it occurred to me that I could use Playback Options → Timing to switch the project back to “straight” and then immediately back to “swing” again and expect Dorico to reset all off-beat notes to the appropriate “swung” start position.

But this doesn’t work. Dorico only re-swings notes which have NOT got a Playback End Offset even though it correctly extends the duration of all the preceding notes, meaning that some pairs of notes then overlap by up to 1/6th of a beat.

When I bought Dorico I’m sure I asked about this and was assured it was as easy as it is in Sibelius but it seems that isn’t correct. I don’t believe it is the intended behaviour that Dorico doesn’t set the swing start position of some notes simpy because you have changed the end position.

To get round this, I will have to manually set a Playback Start Offset for each and every off-beat note that I have set an End Offset for. The value of the offset will depend on whether it is light swing or heavy swing.

This is off-putting in the extreme.

Michael

[FredGUnn]

@Michael_Kilpatrick

In a vast number of cases, off-beat swung notes whether written as a crotchet, two tied quavers, or a quaver tied to a crotchet, are all short notes in most contexts.

What Michael is describing here is the default phrasing in many jazz situations. For years, a quarter note would be assumed to be short in a swing situation, but long in a ballad. Almost all of the Alfred Essential Jazz Editions (Jazz at Lincoln Center) scores are available in perusal form and filled with this type of phrasing. The score to the Jimmie Lunceford version of Avalon is a good example of this (and features a ridiculous alto solo by Willie Smith too). Here’s the audio.

I’m not sure what Ellington stuff Michael is working on, but this was certainly the style in Ellington’s manuscripts as well as all of the David Berger and Essentially Ellington transcriptions. This phrasing is assumed by anyone who knows and plays this style, but obviously is not assumed by Dorico. A setting to play quarter notes short unless marked with a tenuto or under a slur would be very helpful when writing/transcribing arrangements using this style.

OT, for anyone interested, Berger has been doing a cool Zoom workshop where he spends a couple of hours talking down a different Ellington score each month. I’ve been enjoying them anyway. Info here.

[Michael_Kilpatrick]

There’s more to it than Fred suggests. David Berger, who I know, does not produce what you would call urtext editions. As I described above, it is often the case that off-beat notes could be written as a quaver tied to a ccrotchet yet still be intended to be a short note. David edits those notes. I do not and my orchestra and others learn to read and interpret them.

Because Dorico destroys the swing (the start offset) of a note if the user changes the end offset, I have twice as much tedious manual intervention required to get something resembling nice playback.

I can’t see how this isn’t a bug, as the start offset and the end offset are surely intended to be independent. If the the Playback Start Offset for a note remains unset, the swing playback mode should be observed regardless of how much one messes with the duration.

This bug is disappointing to the extent it seems more pleasurable to stick with Sibelius.

[FredGUnn]

@Michael_Kilpatrick

Because Dorico destroys the swing (the start offset) of a note if the user changes the end offset, I have twice as much tedious manual intervention required to get something resembling nice playback.

Another way I think this could potentially be approached is by simply allowing any notational elements that affect playback to be hidden. This has been a feature request of mine for a while now since I’m much more comfortable working with notation rather than MIDI. If the user could add a staccato to any of the relevant quarter notes, then hide the marking, playback would be correct both in terms of duration and also in terms of swing placement. This would be a very useful feature for me anyway.

Perhaps this is too OT for this forum, but I’m very intrigued by what urtext means in the context of an Ellington score. Virtually all of the Ellington manuscripts I’ve seen look like the below concert sketch score with certain players (Rab, etc) individually labelled. Tizol, Tom Whaley, John Sanders, and his other copyists always copied the parts directly from the sketch score. He also never wrote piano or drum parts, and virtually never wrote dynamics. Just curious if your urtext scores reflect these original manuscripts, or if you are using a traditional score layouts, or adding any additional missing elements. Are you doing any from the Irving Mills era? I might be mistaken, but I thought almost all of those original scores were destroyed in a flood or fire (can’t remember which) at Mills’s Florida home. Your urtext project sounds very interesting so I’m curious exactly how you are going about it, and which scores will be included.

[Michael_Kilpatrick]

I’m using urtext in a rather liberal way. I’m merely saying that I don’t adjust any of the written note durations. I do add a bare minimum of dynamics and occasional phrasing. For note durations, the more tied across the middle of the second bar in the top stave is a good example: it is a short note but written as a quaver tied to a crotchet. Berger would edit that. I don’t.

As for the “Mills era”, I think he was manager right up until 1939 or thereabouts? No, Duke Ellington manuscripts from that period were not destroyed in a fire. Most of them exist. Most of them are in the Archives Centre of the Smithsonian Institution. Quite a number of manuscripts from the 1930s are untitled; I have identified quite a few scores and parts including songs that were never actually recorded (by means of knowing beforehand that Duke made arrangements of several dozen popular songs in the early 1930s which are noted in NBC broadcast logs from the Cotton Club).

[Michael_Kilpatrick]

The other thing I do that others (eg Berger) don’t is to ensure that improvised solos appear in the band part as appendices do that in situ, in the relevant section of the musician’s part, the original written melody is shown if it should be. This means, for example, that an alto saxophonist playing “The Star-crossed Lovers” actually gets to see the melody as Strayhorn actually wrote it. Having only Johnny Hodges’ solo, taken by ear from a recording, means that important original information is not present.

[FredGUnn]

@Michael_Kilpatrick

As for the “Mills era”, I think he was manager right up until 1939 or thereabouts? No, Duke Ellington manuscripts from that period were not destroyed in a fire. Most of them exist. Most of them are in the Archives Centre of the Smithsonian Institution.

I might be wrong then. I thought most of those were lost and much of the Smithsonian collection was from the Tempo Music era (including many arrangements of compositions dating from the Mills era), but maybe that’s not right.

@Michael_Kilpatrick

for example, that an alto saxophonist playing “The Star-crossed Lovers” actually gets to see the melody as Strayhorn actually wrote it. Having only Johnny Hodges’ solo, taken by ear from a recording, means that important original information is not present.

I completely agree with this. In the interest of full disclosure, I know Dave well and have even recorded with his band. I’ll add that he usually only does a complete transcription (solos, full rhythm section, etc.) as you described for the Essentially Ellington stuff, as that is what Jazz at Lincoln Center wants for publication. For his own band and transcriptions for other professional groups he usually writes improvised sections as slashes, and melodies distilled down to their essence to be embellished. Different requirements for different markets I guess.

Back on the Dorico front, if there was a way to simply add a staccato to any of the short quarters, then hide them, do you think that would be helpful for your work? I know it would be a useful feature for me. Alternatively, you can always have one score for publication, and one for playback if required. Go through and mark the short quarters with a staccato in your “playback score” and they should still swing correctly, eliminating quite a bit of the tedious MIDI work.

[dspreadbury]

Thanks for your feedback on this issue, Michael. The current behaviour is certainly intentional, but quite possibly it’s more restrictive than it needs to be, so we will look into relaxing the limitation that a note can have neither a custom start or end offset to be considered for swing.

I’m interested to learn whether there is something we could do automatically to make it unnecessary for you to adjust the duration of off-beat notes at all. Is there a simple (or simple-ish) rule that governs the adjustment you are making manually?

[Michael_Kilpatrick]

Your idea of invisible staccatos would help to some extent but it isn’t just crotchets that need shortening and the extent of the shortening isn’t the same for each. So a single, simple mechanism would not be capable of achieving the total result without a set of parameters and options.

Ideally the first step would be for Dorico just to do what I expect it to go by setting a Playback End Offset; it would not ruin the Start Offset as defined by the swing.

In the future I would ike a clever swing playback option which would automatically shorten crotchets to a certain extent and also, optionally, shorten other note durations under my control with a set of parameters. This would minimise the extent of manual intervention through Playback End Offsets which are a bit of a drag. This more clever playback could be systematic and I’m pretty sure I could define it in a set of parameters.

I would also like to control the extent to which the last note under a slur is shortened too, variable according to its written duration.

[Michael_Kilpatrick]

Daniel, I must say I’m baffled by your saying the current behavior is intentional. The swing playback defines a start offset for all offbeat notes. When the user has not specified manually a Start Offset it seems utterly counter-intuitive that the swing offset would be eradicated because of the presence of an End Offset.

Edit:you say "restrictive’ but what it does is break the swing playback. To resolve this issue one has to manually enter a Start Offset. In order to do that one has to know what the correct value of Start Offset is required depending on light/medium/heavy swing. I would say it’s broken.

[FredGUnn]

@dspreadbury

I’m interested to learn whether there is something we could do automatically to make it unnecessary for you to adjust the duration of off-beat notes at all. Is there a simple (or simple-ish) rule that governs the adjustment you are making manually?

I don’t want to speak for Michael here, but if you look at and listen to the trombone soli at letter F in the Avalon score I linked to above, Trombone 1 has:

This is the standard way this would be written in this era, but would in fact be played like below (with a swing feeling) by anyone used to reading this style:

Michael may need more sophisticated adjustments than simply that, but the ability to have the quarters play back short and still swing would be very useful for me anyway!

[pianoleo]

@FredGUnn

Back on the Dorico front, if there was a way to simply add a staccato to any of the short quarters, then hide them, do you think that would be helpful for your work? I know it would be a useful feature for me.

This is already possible. Create a Playing Technique that triggers the Staccato Playback technique, like so:

Apply it where needed.

Hide from properties panel:

[FredGUnn]

This is a really good solution, thanks! Staccatissimo is virtually unused in this style so if Michael needs another subtle gradation of the “Play as %” setting, that would be one other additional option too (that could be longer or shorter as defined).

[Michael_Kilpatrick]

Thanks. That’s a, solution that can work but only for notes you want to reduce by a fixed amount equivalent to that staccato. Unfortunately my needs are much wider than that. Having Dorico not destroy the swing placement when using the Playback End Offset would be a better solution, perhaps followed later by some more sophisticated options for playback.

[pianoleo]

@Michael_Kilpatrick

To get round this, I will have to manually set a Playback Start Offset for each and every off-beat note that I have set an End Offset for. The value of the offset will depend on whether it is light swing or heavy swing.

By the way - and I’m not disagreeing with your feature request in the slightest, only aiming to make your life easier while things as they are - you don’t need to manually calculate the Playback Start Offset. Just toggle the Playback Start Offset switch on and then off again.

You could even Select All, Filter Notes and then toggle the switch on and off once for the whole flow.

[Michael_Kilpatrick]

There is a problem, I think, in that if you just select all or indeed select any large passage in which there are a few notes for which you have already assigned a Playback Start Offset for some reason, you can’t simply toggle on and off again. You have to be sure to select only notes with no manually assigned Offset.