virtual mock-up vs final score

[michael_maberly]

I would really welcome any input from those of you who have experience with this problem:

For years, I used midi strictly as a sketch-pad prior to writing a real (classically-oriented) score for real (classical) players. Then I discovered higher-end orchestral samples, got tired of waiting for easy access to good (real) orchestras :rolleyes: , and decided to put my heart into my virtual scores first.

Although I often sketch my lines and parts out on paper beforehand, I now enter them via my midi keyboard. As a result, they now breathe better, and more often than not possess the magic that can happen in real performance (at least in the terms of time).

But here's the problem (as I'm sure you've already anticipated): the magic often lies between the cracks of conventional/acceptable-to-read notation, and rendering the parts into user-friendly notation often destroys the sense and meaning of the more subtle midi performance (again, in terms of time).

Here's an example: Over a four-part texture that was written in strict quarters, eighths and sixteenths, I played in two more parts, each moving outside of the meter and around the beats, to create a kind of harmonically moving, free-floating, two-note cluster. I was delighted with the over-all effect: it captured the very feeling I'd been looking for.

When it came time to consider how it would look in a real score, I took a look at every note's Start Position, in terms of bars/beats/16ths/ticks (I use Cubase SX3). Of course, none of them lined up with the usual subdivisions of the beat, and none of them lined up with other bar- and beat-subdivisions (like quintuplets, septuplets, etc). I naively rounded everything off to the nearest main subdivision (16ths or 8th triplets), and the result was a complete disappointment. I then spent many hours experimenting with rounding everything off to finer subdivisions -- 32nd notes, and 16th-note triplets -- always trying to keep the proportional relations between the two parts (and their relation to the 'fixed' parts) close to the original. The final result was something that's still not easy to read, and that only hints at the original's magic.

How do you approach this problem -- that is, the problem of translating the timing of a midi-keyboard 'performance' into the fixed and rigid requirements of a (classical) score, for classical musicians?

Thanks, in advance, for sharing your perspectives on this!

[joaz]

The answer lies in having 2 versions.
1 Sounds Good
2 Looks good.

Do the Sounds-Good one first, and then quantise it till it looks good.

regards Joe

[michael_maberly]

The answer lies in having 2 versions.
1 Sounds Good
2 Looks good.

Do the Sounds-Good one first, and then quantise it till it looks good.

regards Joe

Thanks for the feedback, Joe.
If it were simply a case of creating a respectable-looking score that isn't going to be played, I'd have no problem! The problem is that the final score is to be performed, by real musicians, and that seems to mean that either it won't sound like the original, or, it will be impossible for flesh-and-blood musicians to read it. I'd love to hear your (and others') perspectives on dealing with thatproblem.

[joaz]

Thanks for the feedback, Joe.
If it were simply a case of creating a respectable-looking score that isn't going to be played, I'd have no problem! The problem is that the final score is to be performed, by real musicians, and that seems to mean that either it won't sound like the original, or, it will be impossible for flesh-and-blood musicians to read it. I'd love to hear your (and others') perspectives on dealing with thatproblem.

I came across this problem recently when writing a Flute Sonata.
It was in a Brasillian Jazz style, and I attempted to notate the real rhytms that Jazz musicians actually play.
I got it to a stage that it sounded right, but of course it was unreadable, because it is essentially impossible to notate Rubato.

The only solution is to get it as close to what you want, while keeping it readable, and then trust your musicians.

I have posted a couple of pieces with scores to view here.
http://www.northernsounds.com/forum/showthread.php?t=44310

I think this is one of those questions that does not have a perfect answer.
Just different degrees of compromise.

regards Joe

[snorlax]

I would really welcome any input from those of you who have experience with this problem:

For years, I used midi strictly as a sketch-pad prior to writing a real (classically-oriented) score for real (classical) players. Then I discovered higher-end orchestral samples, got tired of waiting for easy access to good (real) orchestras :rolleyes: , and decided to put my heart into my virtual scores first.

Although I often sketch my lines and parts out on paper beforehand, I now enter them via my midi keyboard. As a result, they now breathe better, and more often than not possess the magic that can happen in real performance (at least in the terms of time).

But here's the problem (as I'm sure you've already anticipated): the magic often lies between the cracks of conventional/acceptable-to-read notation, and rendering the parts into user-friendly notation often destroys the sense and meaning of the more subtle midi performance (again, in terms of time).

Here's an example: Over a four-part texture that was written in strict quarters, eighths and sixteenths, I played in two more parts, each moving outside of the meter and around the beats, to create a kind of harmonically moving, free-floating, two-note cluster. I was delighted with the over-all effect: it captured the very feeling I'd been looking for.

When it came time to consider how it would look in a real score, I took a look at every note's Start Position, in terms of bars/beats/16ths/ticks (I use Cubase SX3). Of course, none of them lined up with the usual subdivisions of the beat, and none of them lined up with other bar- and beat-subdivisions (like quintuplets, septuplets, etc). I naively rounded everything off to the nearest main subdivision (16ths or 8th triplets), and the result was a complete disappointment. I then spent many hours experimenting with rounding everything off to finer subdivisions -- 32nd notes, and 16th-note triplets -- always trying to keep the proportional relations between the two parts (and their relation to the 'fixed' parts) close to the original. The final result was something that's still not easy to read, and that only hints at the original's magic.

How do you approach this problem -- that is, the problem of translating the timing of a midi-keyboard 'performance' into the fixed and rigid requirements of a (classical) score, for classical musicians?

Thanks, in advance, for sharing your perspectives on this!

Michael, I am in your same boat, or close to it. Several possibilities. You didn't state which notation package you use, so I'll comment on the two I know and use regularly: Overture 4 and Finale 2006.

One technique is to use the "retranscribe" function in either program. Without SEEING what you wrote, I can say with some confidence that "retranscribe" will make your music look like it's supposed to and play back as it's supposed to. Simply highlight a passage and do the retranscribe thing in either program, selecting the note value that makes most sense for the passage.

Additionally, different sections of a piece may require different quantization settings. If you're not doing so already, you might enter a passage using one quant. setting, then stop the live entry session, change the quant settings, then enter the next section.

Additionally, both programs have programmable swing percentages if that is a concern. Both also offer a degree of humanization--randomization of start time, stop time, and note length. I don't know if FinaleScript can operate at theat level or not.

The consequence there is that I often enter the music in step-time, square, and let Overture provide the swing and humanization.

Just some before-dinner thoughts...when I'm nourished after a day of brass quintet gigs, I'll add some more thoughts.

Jim

[michael_maberly]

I came across this problem recently when writing a Flute Sonata.
It was in a Brasillian Jazz style, and I attempted to notate the real rhytms that Jazz musicians actually play.
I got it to a stage that it sounded right, but of course it was unreadable, because it is essentially impossible to notate Rubato.

The only solution is to get it as close to what you want, while keeping it readable, and then trust your musicians.

I have posted a couple of pieces with scores to view here.
http://www.northernsounds.com/forum/showthread.php?t=44310

I think this is one of those questions that does not have a perfect answer.
Just different degrees of compromise.

regards Joe

Thanks, both Joe and Jim -- I appreciate your input.
Joe, I'm downloading your quintet mp3 & score as I write -- I'm on a slow dial-up, so it'll take about an hour and a half!

I think - alas! - that you're probably right about different degrees of compromise. However, I also think the problems and solutions are different in classical and non-classical music: anyone who plays Brasilian music will know the rhythmic approaches it demands, and will most likely interpret the score accordingly. Swing or bop players will know to interpret eighths as swing eighths, unless otherwise directed, and blues guitarists will know when and how to bend notes, whether that's spelled out or not. But -- with the exception of their own customs, like vibrato, etc-- classical players (I think) try to avoid adding what isn't there. So the problem I see is to either make the reading so difficult that they'll either play it incorrectly (or, more likely, just not play the piece), or make it playable but incorrect.

Jim, I can see how the re-transcribe function in Finale or Sibelius could save some time, but -- as you can probably see from the above -- my problem isn't in how to write what I've played (although it is time-consuming -- excuse the pun), but in getting any rational human being to put up with trying to play it :) . The question of what level of quantization to use seems to be the same, whether in a notation program or sequencer: the score will be much closer to the effect I've played if I use 64ths and all combinations of tuplets, but it'll drive the poor performers crazy, and either won't get played properly, or just won't get played!

It's a real conundrum, especially when something sounds so good as played in!

[michael_maberly]

I have posted a couple of pieces with scores to view here.
http://www.northernsounds.com/forum/showthread.php?t=44310


Finally downloaded your quintet, Joe -- very nice!! I hope you'll be around on this forum for awhile yet: I'd like to take a close look at your piece when I get some time, and will probably have some questions for you!
By the way, I only realized the PDF was a C score when I finally downloaded the MP3!

All the best,
Michael

[Rach3]

There have been contemporary composers writing ridiculously complicated rhythms, hoping to get full control of every details of the piece. In reality, it's just not possible. Every human person has a different "feel" for interpretation. You can't really "force" your own interpretation onto your performer 100%. The best way to do it, IMO, is to leave some space for the performers to breathe in their own interpretation. A lot of performers hate reading a score cluttered with many annoying details.

Solutions to your problem:
I would probably keep the score as simple as possible, and give them the demo to listen to. Also, try to explain to them verbally what you require. And most importantly, don't try to control everything.

[JonP]

There have been contemporary composers writing ridiculously complicated rhythms, hoping to get full control of every details of the piece. In reality, it's just not possible.

I think that's very true. I remember listening to a quartet by Ferneyhough when I was a student and then we were asked to notate, by ear, what we heard. It was only a few seconds long but the amount going on was bewildering. Our professor then produced the score and it looked pretty different to what we had written. But, perhaps suprisingly, a lot of us had written it down more accurately than Ferneyhough. It didn't take a detective to work out that the performers weren't playing what was written....not by quite a margin and yet the composer had endorsed the recording on the sleeve notes as being a great performance.

So did the composer have not a clue as to what he'd written?. Or was he trying to obtain a sense of chaos by purposely convoluted notation?. Was the precision of the performance not as important as the notation seemed to demand?. Could Ferneyhough have created the same results with a more
aleatorically centered notation system?. If I think I am a cloud, will I start to float? (....sorry, wrong forum). It was a very revealing lesson.

BTW, I'm with Joaz 100%. Trust your performers.....excellent advice.

[michael_maberly]

Thanks, Rach3 & JonP.

Jon, I've witnessed similar things -- I recall being invited to look at the score an avante-garde string quartet had just performed, and being astonished at how different it was (rhythmically) from what was written -- and it wasn't an aleatoric composition! I have good friends who bemoan the inaccuracy of the few orchestral performances they get. And I'm aware of how aleatoric composition arose, in part, as one way of achieving certain textures without demanding the impossible of performers.

I agree, Rach3, that one's more likely to get an expressive performance when the performer's attention isn't being completely absorbed by impossible minutae. In the particular passage I described (above), though, there is no room for any variance from what was originally played: I spent many hours re-recording the same passage from easier-to-read notations, only to find that none was able to recreate the original's magic. (And an aleatoric approach wouldn't have worked, either.) It wasn't a question of an expressive solo melody, but of clusters ornamenting a contrapuntal passage. I guess I must have accidentally stumbled on some micro-rhythms that created the magical effect -- micro-rhythms I still don't understand: how is it possible that a few quintuplets, septuplets and 64ths -- as opposed to 16th-note triplets and 32nds (quarter beat = 80) could make such an audible -- and powerfully emotive -- difference?! In the end - since there's no way human performers could accurately duplicate the passage as originally written - I have to settle for something that's really half-baked.

I would think that the problem of mock-up vs final score would be only too-well known by sample/midi-composers, who probably have more experience with this type of problem -- and more need of solutions --than most. For most of us enter our notes via midi keyboard, and then -- if we want our work to be re-created by real performers - have to surrender our compositional and performance intricacies to the limitations imposed by notation.

[joaz]

Michael

I totally understand the fact that, when playing, we can stumble across magic rhythms.
However, when preparing scores for performance, we are limited to notating understandable magic.
There is a trade off involved, because, the performers will hopefully bring some magic of their own, which we didn't put in the score.

Imagine trying to take down by dictation, a Chopin waltz played with a high degree of Rubato. It would probably end up looking like a Fernyhough score.

I have always believed that a score contains about 50% of the information, required to pull off a convincing performance. The rest is what the performers bring to the party.

If you are content to let your music exist in a private virtual universe, then you can happilly enjoy your magic rhythms.
If you want it out in the world, being played by flesh and blood musicians, then a degree of compromise is necessary, I fear.

regards Joe

[michael_maberly]

Imagine trying to take down by dictation, a Chopin waltz played with a high degree of Rubato. It would probably end up looking like a Fernyhough score.

Hi, Joe;

This is an interesting point, that makes me wonder what stylistic innovators of the past had to do to obtain truthful orchestral performances -- that is, before their styles became widely accepted. Certainly, instrumental composers, like Chopin, Liszt, and Paganini stimulated emulation by virtue of their own spell-binding performances. But when it came to their orchestral work, this facet (rubato) necessarily remained limited to concerti -- where only the soloist could indulge. Perhaps only the 20th century demanded orchestral virtuosity -- and the question of whether it actually received it is still valid.

[sinkd]

Hi Michael,

This is an interesting question--one that composers have to deal with a lot. One suggestion I have is to get the score as close as possible using tuplets, rubato arrows to show "laying back" on the beat, or whatever you need to do. Then, deliver a good audio mockup to the players who are going to perform it. If you have a website, you could also permanently post this and put the link right on the score.

I have found some performers very willing to be "coached" by mockups, some aren't interested. My feeling is, all we are doing as composers is trying to communicate the sounds in our head to the performers on stage. MIDI scores, sample libs, etc. are all just new parts of what we can call "the score."

DS

[jsp2]

This is an issue I've been wondering of for quite a while.

I've never intended for my work to be performed, so it's not a major concern, however it would be nice to see an accurate visual notation regardless.

You mentioned you were using Cubase SX3... I too am using Cubase SL, There's a function called "Time Warp, or "Stretch", (?) which allows you to keep your midi information in tact, while adjusting your Tempo track.

I've only used if briefly, so I would not be able to fully help with it's functionality, But this may be a direction to look into.

There will no doubt be some manual adjusting, but it could be a good place to start.

The Cubase Forums may be a good place for this information.

~Jeff

[snorlax]

Friends...

Much of the discussion seems to center around accels, rits, and the like. Please remember that both Finale and Overture 4 have TEMPO TAP capability for this exact purpose.

It may be the case that no ONE tool can give 100% of the desired mix of correct appearance and correct playback, but consider the possibility of different entry methods (step in some places, realtime in others), tempo tap, quantizing, humanizing, human playback, etc. If we apply several tools, perhaps we can come closer to the goal. Also remember that it doesn't ALL have to be done upopn entry. Much of it can be post-notation as well.

Especially in Finale, the apply HP plugin may be quite useful for a given segment of a score.

[michael_maberly]

This is an interesting question--one that composers have to deal with a lot. One suggestion I have is to get the score as close as possible using tuplets, rubato arrows to show "laying back" on the beat, or whatever you need to do. Then, deliver a good audio mockup to the players who are going to perform it. If you have a website, you could also permanently post this and put the link right on the score. DS

Good suggestions, Damon -- thank-you. In terms of getting the score as close as possible, I wonder what you (and others) think of this: I can get the parts -- which are in 3/4 time, with quarter note = 80 -- to 99% accuracy by notating them as a series of quintuplets, septuplets, 32nds, 9-tuplets and 10-tuplets (my Latin ends at "seven"). However, I doubt that many players could play them, even if they wanted to. A toned-down version limits the parts to 16ths, 16th-note triplets, and 32nds -- which might be playable, but which doesn't create the same effect. Any suggestions as to more intuitive ways of conveying the intended effect?

I hadn't heard of rubato "arrows" before. Can you describe them?

Thanks for your interest in this thread!

[michael_maberly]

You mentioned you were using Cubase SX3... I too am using Cubase SL, There's a function called "Time Warp, or "Stretch", (?) which allows you to keep your midi information in tact, while adjusting your Tempo track.
Thanks, Jeff. Although I'm not quite sure of this, I think Time Warp creates a correspondence between linear time (in seconds) and musical time (in beats).
Actually, I've a very accurate way of figuring out difficult quantization/notation, and that is by using the Start, End and Duration data in SX's Info Line (the Info Line sits just above the editing window -- click on the "Show Event Info Line" icon, which is just to the right of the "Inspector Icon", to display it). The data is shown in the "bar/beat/16th-note-position-within-the-beat/ticks" format. For example, let's say a note has a "Start" number of "1.2.2.27": this means that it begins at "bar 1,beat 2, after the first 16th of beat 2, plus 27 ticks". Since SX divides every quarter note into 480 ticks, this means the note begins 147 ticks into beat 2 (a 16th = 120 ticks, + 27 ticks). I then try to figure out what kind of time value or tuplet comes closest to 147/480. It's very mechanical and tedious, but is accurate and reliable, when necessary (I wouldn't bother with it for routine quantization!).

[Houston Haynes]

How do you approach this problem -- that is, the problem of translating the timing of a midi-keyboard 'performance' into the fixed and rigid requirements of a (classical) score, for classical musicians?

Thanks, in advance, for sharing your perspectives on this!

Use Time Warp (ther's more than one mode) and Tempo mapping to align your bars and beats as close as you can, then use Quantize display in the SX3's Score Editor. Problem solved.

It sounds like you could use a few lessons from Cubase SX Unleashed (http://www.auxbuss.com/products/sxu/) - money well spent.

[michael_maberly]

Use Time Warp (ther's more than one mode) and Tempo mapping to align your bars and beats as close as you can, then use Quantize display in the SX3's Score Editor. Problem solved.

It sounds like you could use a few lessons from Cubase SX Unleashed (http://www.auxbuss.com/products/sxu/) - money well spent.

Thanks, Houston -- I'll take a closer look at Time Warp. But I wonder if I'm missing something: my problem isn't in how to accurately notate what I've played (see above), but in what to do with an accurate notation that is probably impossible for a live musician to accurately play. I'm not sure I understand how Time Warp can help me with that.

[Houston Haynes]

Heh. If I've got an accurate part that's important and I can't find a player to pull it off and *add* valid interpretive elements that I cannot - then I keep the part I've recorded - happens all the time. Again, problem solved. Harry Gregson-Williams spoke openly about keeping the virtual percussion parts for his score to "The Lion, The Witch, and The Wardrobe". His point was more about accuracy and holding onto the beat, but it's essentially the same issue.

Otherwise - get it musically close on paper and give them an audio clip to listen to... a great example of this is Chick Corea's "Spain" - two distinct versions are popular, and the while the notation is different for each, they're still both woefully short on what you really need to do in order to pull off the lick. That's life.

Keep it simple. If you can't write it - then get it close and give someone a shot at the rendered part and let them tell you how they'd like to see it - or just let them pick it up by ear. If there's a better way to notate it such that they can more effectively transmit your intent, they'll give you suggestions on how to do that if they're worth their salt. Also, study other scores that have crazy parts in it - there are many times where I've heard something and thought "how the HELL did they get that on paper?" only to look at a study score and say "you got THAT from THAT??" That's life, too.

In the end, music notation is just a two-dimentional representation of something that only correctly exists in four dimensions - and even then only in the hands of the properly qualified. That's some pretty rare air - don't try to shoe-horn music notation into something it's not meant to do.

[Von Richter]

If you do many things beyond the limits of a real instrument, you might consider including actual audio to be triggered as part of the ensemble. I wish this were more common outside of recording. It has amazing potential live.

[Houston Haynes]

Mario Davidovsky's "Synchronisms No. 1" for flute and electronic sound.

[michael_maberly]

Mario Davidovsky's "Synchronisms No. 1" for flute and electronic sound.
Thanks Houston, for all your comments and suggestions, and thanks Von, for bringing up the idea of integrating actual audio with the live performance. I know that many composers have used that technique, but have never tried it myself. Any suggestions on the best way/format for "triggering" it accurately within a concert performance (ie, as one small audio part within a live orchestral performance)?

[Hermitage59]

If you do many things beyond the limits of a real instrument, you might consider including actual audio to be triggered as part of the ensemble. I wish this were more common outside of recording. It has amazing potential live.

And the epic 'African S a n c t u s.' A highly integrated piece that was performed some years ago, with orchestra and large tape deck. The composers name is David Fanshawe. Interesting work. (Took ten years and thousands of miles of travelling to record the natural voices and music). I think the mixer was the chap with the hardest job, as much of it relied on the harmonic overtones created by natural dissonance in the hall space (Sydney Opera House Concert Hall), and he had to get it just right. (Which he did.)
I would think the current digital tools would be ideal for this grand scale sort of work. I'm not sure if the piece is still in print (Audio), but it was an interesting experience being part of an ensemble playing this.


regards,

Alex.

[Hermitage59]

Thanks Houston, for all your comments and suggestions, and thanks Von, for bringing up the idea of integrating actual audio with the live performance. I know that many composers have used that technique, but have never tried it myself. Any suggestions on the best way/format for "triggering" it accurately within a concert performance (ie, as one small audio part within a live orchestral performance)?

The simplest way, and i've done this once or twice, is to treat it as an instrument, and assign a part to the 'player.' He then aligns the audio, (with a precise lead in time if neccessary to offset any chance of initial stutter)
and performs.
I say this from the experience of working with tape decks (Grundig reel to reel with integrated computer sync. Still the finest 'hardware' i ever had the privilege to own, and the 'warmest' recordings i ever made), so i'm not that familiar with the multimedia integrative techniques used currently.


Alex.

[michael_maberly]

The simplest way, and i've done this once or twice, is to treat it as an instrument, and assign a part to the 'player.' He then aligns the audio, (with a precise lead in time if neccessary to offset any chance of initial stutter) and performs
Thanks, Alex.

[CallMeZoot]

I run into this kind of problem all the time.

I do my mocking up in SONAR, before I even think about scoring. Don't let your music be limited by the limitations of notation -- write what you want in your sequencing software (if you use sequencing software). When you are satisfied with the sound of your mockup, you have your piece. THEN it's up to you and your musical training (and the advice of musician friends) to figure out the best way to communicate it on paper. Once you've mocked up your piece in a sequencer and it sounds the way you want, export it into your notation software. Do all the quantizing and tweaking you want, but NEVER HIT PLAY in your notation software or you will start to second guess yourself.

Sometimes it's just a matter of writing out a complex rhythm, tuplet or whatever. Sometimes your computer gets it right, it just doesn't SOUND right when it's played by a computer. If you suspect this is the case, try reading the part (or better yet, show it to another player), and playing it yourself. You'd be surprised how the same rhythm might sound right when it's channeled through a human brain, even if a computer's "accurate" rhythm. I don't suggest writing out really crazy complex rhythms, but if you are working with pro players who are used to contemporary music, they can figure out a lot of crazy stuff.

In cases where no notation yields a satisfying result, congratulations! You've succeeded in writing music that is more human than dots on paper could capture. This is not your error, this is the limitations of the language.

In this case you might want to check with a player and ask how they would prefer it to be notated. Or even try a few different options and have a player sight-read all of them -- see which comes closest to what you want, and go with that one.

Don't underestimate the value of TEXT in your scores. In your case you might just want to write stemless quarter notes and write "out of rhythm, behind beat" or something. Most players could figure that out just fine. If you want to get more specific, you could place the noteheads graphically, in relation to vertical lines or hashes which demonstrate where the beat is. This way the player can see -- this note should come just a bit before this beat, this note should come just a bit after this beat, etc. That will get them pretty darn close without requiring them to count some crazy nested 32nd triplet figure.

I would recommend two books in particular. Both have titles along the line of "Techniques of the Contemporary Composer" or something like that. One is by Vincent Persichetti, and covers a lot of 20th century techniques. Another is by David Cope, and covers more modern, avant garde techniques. These might give you a lot of ideas for notating complex, non-traditional ideas.

chris.

[michael_maberly]

Thanks for all your suggestions, Chris!
I've had the Persechetti for about thirty years (!). I think I've seen the Cope, but can't recall its contents.