Posted
We have discussed chromatic alterations and how they would impact the harmonic context etc. I don't really see the necessity of that approach.
I would like to see a parameter for altering the output midi notes by arbitrary scale steps. It would not, and should not be considered in the harmonic context at all (the harmonic context would be looking at everything before the shifts)
I am coming at this from an orchestral scoring perspective but I think it applies across the board.
In orchestrating parts it is common to double instruments at a constant interval. An example of this is soft trumpets doubled at a constant Major second; the effect in that case is not dissonant. Like wise synth patches with a pair of oscillators tuned to a constant perfect 5th are quite common.
In most cases the effect of this kind of doubling is not structural but for effect and timbre. They shouldn't really be thought of as a separate harmonic layer.
If a chromatic parameter was added parts can be doubled and, more importantly, updated and edited, without mainting two separate figures and harmonic layers.
This is also the only way to guarantee that synfire maintains the doubled intervallic relationship no matter what, without resorting to the dreaded [p] symbols.
I imagine coding such a Chromatic parameter would be fairly easy as it would intently NOT interact with the rest of the environment.
What do you think?
Thu, 2009-08-27 - 09:50 Permalink
Sounds like an interesting approach. Any concerns or suggestions from others?
Fri, 2009-08-28 - 18:43 Permalink
Another example where this would be useful would be a jazz pianist wanting to voice a chord in 4ths - ie play a chord with two 4ths below the melody note.
A similar problem is encountered by that same pianist wanting to guarantee that the 3rd and 7th are played in the left hand. I still haven't found a way of doing that easily in Synfire.
I guess we're talking about chord voicings. I think it's a very important area. The scaleboard and the whole display gives the impression that synfire is going to stick with the same scale steps when you duplicate a figure into a different harmony but it doesn't. I realise that normally you might want the current behaviour but anyone concentrating on an arrangement might well want to keep the same scale steps ....
So yes please - a way of "freezing" or specifying the intervals in chords would be wonderful!
Fri, 2009-08-28 - 20:17 Permalink
Hi Tim,
I think the area that your discussing is a separate issue. Though, I agree I would like more control in that realm as well.
What I'm looking for is a parameter that solidly declares itself outside any harmonic considerations.
I know Synfire has certain conceptual goals as far as harmony is concerned. I imagine any changes in that realm must satisfy a great number of concerns and exceptions.
A chromatic shift parameter is something that would be very easy to add and would not break any of the conceptual thinking currently implemented into Synfires harmonic context.
The goal is to use chromatic shifts relative to an understood tonal harmony. It is not linked to a particular chord or voicing. Synfires harmonic and structural flexibility remains unaffected.
Chords can change, new phrases can be recorded on other instruments and still effect the voicing of the chromatic part -- just always with a +1 or -5 etc.
Sat, 2009-08-29 - 18:08 Permalink
Hi Tokyorose,
Yes - I was actually making two requests.
First a way to tell synfire that a note should always be a set number of semitones apart from another. If you then copy and pasted those notes you'd end up with what I know as "constant structures" which occur a lot in band arranging as well as, as you say, orchestral works. I think that similar to your request but maybe not quite the same?
My second request is to be able to tell synfire that a particular note should always be a particular scale step of the current chord. That's definitely different to your request!
Sat, 2009-08-29 - 19:05 Permalink
This topic is definitely an important one that deserves a discussion. We much appreciate your ideas and suggestions, thank you so far.
There are multiple possible ways to implement some form of chromaticism or static intervals or "out of scale" melody or chord alterations (and: chromatic passages, like walking bass or successive chords wandering up chromatically).
One that has not been mentioned yet, and which is probably the most obvious, is using the figure symbols directly. They already have a chromatic pitch component that could be used to express chromatic intervals measured from a scale step. We only need to tell the renderer to take these literally instead of rounding them up/down to valid scale steps of the moment. Perhaps a simple segment property or interpretation switch would be sufficient.
This way a chord can be expressed as a root (zero line) with only chromatic intervals. That is, all symbols have the same zero diatonic component but different chromatic shifts.
The chord (<0,0> <2,0> <4,0>) might become (<0,0> <0,+3> <0,+6>) depending on context. Some command may support easy conversion in the current context, hmm, "Chromaticize Chord Intervals" or something.
I am somewhat reluctant to introducing a new parameter for this, if it can be done with the figure alone.
The top priority rule of music prototyping is: Any set of figures that work together in one context should also work in any other context. Breaking this rule means losing reusability.
My biggest concern is if and how manual chromaticisms will violate the overall harmonic integrity. How much static intervals can a figure bear without conflicting too much with the other instruments when a different harmonic context is applied (IOW: How awful will it sound with another progression)?
This can only be estimated by experimentation. I should probably add a couple hidden experimental features and we can do some experiments together and share our findings.
Sat, 2009-08-29 - 19:28 Permalink
[quote="Tim UK"]First a way to tell synfire that a note should always be a set number of semitones apart from another. If you then copy and pasted those notes you'd end up with what I know as "constant structures" which occur a lot in band arranging as well as, as you say, orchestral works.
An interesting question is whether these constant structures take over the role of altering harmony as a whole (temporarily, valid for all instruments), or if that is really only melodic variety of a single instrument or group that adds to the current harmony.
This by no means a philosophical question. It's crucial for the implementation in Synfire. :?
[quote="Tim UK"]My second request is to be able to tell synfire that a particular note should always be a particular scale step of the current chord.
The green symbols do exactly this, provided by "scale steps of the chord" you think of the vertical scale. Or you could use the purple arpeggio symbols if you want only the notes of the chord alone, the advantage being that these notes are always available, regardless of voice leading constraints.
As one of the experimental features, I could add a switch to the segment properties that disables voice leading for that segment, so the entire scale will be reliably available.
Sat, 2009-08-29 - 21:48 Permalink
This topic is definitely an important one that deserves a discussion. We much appreciate your ideas and suggestions, thank you so far.There are multiple possible ways to implement some form of chromaticism or static intervals or "out of scale" melody or chord alterations (and: chromatic passages, like walking bass or successive chords wandering up chromatically).
One that has not been mentioned yet, and which is probably the most obvious, is using the figure symbols directly. They already have a chromatic pitch component that could be used to express chromatic intervals measured from a scale step. We only need to tell the renderer to take these literally instead of rounding them up/down to valid scale steps of the moment. Perhaps a simple segment property or interpretation switch would be sufficient.
FigureA has a number of symbols with their typical rounded chromatic component. A plain vanilla typical Synfire figure.
I want to take figure A and make a duplicate of it a minor 2nd up -- figure B.
I want the figure B's symbols rounded in the exact same way as figure A based on the harmony -- an exact duplicate. I then want to shift that entire segment some direction by a constant interval. Lets say, a minor 2nd...
I am likely going to be replacing figure B MANY times -- every time I make any adjustment to figure A I will be copy/pasting it over figure B. I want the two to match exactly. (Id rather have segment based aliases, which I would love to discuss but for the moment...)
I don't wan't to have to run any functions on figure B every time I update A, copy/pasting every two seconds is a pain as it is.
The progression can change any which way. Figure A will always fit the progression as per the rules of Synfire. Figure B will always match figure A but with a minor 2nd offset.
I NEED Synfire to round the chromatic symbols of B in exactly the same way as A.
If I decided that I want a Major 3rd instead of a minor second, I can change that in a heart beat.
The interval of the doubling can be organized structurally and clearly displayed as a parameter (not hidden from comprehension in the chromatic shifts of a figure).
This way a chord can be expressed as a root (zero line) with only chromatic intervals. That is, all symbols have the same zero diatonic component but different chromatic shifts.The chord (<0,0> <2,0> <4,0>) might become (<0,0> <0,+3> <0,+6>) depending on context. Some command may support easy conversion in the current context, hmm, "Chromaticize Chord Intervals" or something.
I am somewhat reluctant to introducing a new parameter for this, if it can be done with the figure alone.
The top priority rule of music prototyping is: Any set of figures that work together in one context should also work in any other context. Breaking this rule means losing reusability.
This is the issue. I don't want to do ANYTHING to the figure. For a long phrase, manually adjusting the chromatic components of individual symbols (those adjustments depending entirely on the current harmonic context) would bring any project to an absolute halt. Then redoing it every time the original changed?
The moment the harmony changes all of the shifts would have to be redone.
With the proposed approach, the task could be done in a matter of seconds.
Changes in the harmonic context would be automatically taken into account with out any need for further adjustment.
My biggest concern is if and how manual chromaticisms will violate the overall harmonic integrity. How much static intervals can a figure bear without conflicting too much with the other instruments when a different harmonic context is applied (IOW: How awful will it sound with another progression)?
This can only be estimated by experimentation. I should probably add a couple hidden experimental features and we can do some experiments together and share our findings.
With a shift parameter, manual chromaticisms would be 100% portable to any other Harmony.
This is a conscious decision... If it sounds awful, it is by our will that it does.
Keeping it in a separate parameter makes it very obvious what is happening and why.
Again, this in not a general solution to chromaticism in Synfire, though It may be used as such until we get something better.
This is a way to quickly and structurally deal with intervalic doubling.
If this breaks reusability, what am I not under standing?
Sat, 2009-08-29 - 22:40 Permalink
[quote]This is a way to quickly and structurally deal with intervalic doubling. Absolutely, but it is only that. It does not help with the other chromaticisms:
A) Chromatic chord movements
B) Chromatic melody alterations (blue notes, walking bass)
C) Precise definition of intervals between independent segments, possibly even across different instruments (which is what I understood Tim was asking for)
For me (A) clearly indicates a change in harmony, i.e. the desired chords have to be worked into the progression. There is no limit as to how wild or quick these can be. This also has the advantage that all instruments nicely respect the changes and, as long as the global key is retained, find their way across that "bridge" in a natural fashion, melody-wise.
Case (B) might be more easily expressed through the figure as explained above. A walking bass segment might have the anchor at a certain chord interval and move to/from that in chromatic steps.
The most difficult case is (C) when used across different instruments, because there are so many other parameters that influence the final result: Playing ranges, interpretation settings, etc.
The simple "Chromatic" parameter is no problem, technically (basically it's just midi postprocessing like Shift). I would however prefer to know whether there are at least a few more applications beyond intervallic doubling. Any ideas?
BTW: A clean way to ensure that narrow voices to not cross and maintain certain intervals is to group them in to a segment (see attached picture). This does not work across instruments, though.
This is a single segment (two_voices2.png)
Sat, 2009-08-29 - 23:33 Permalink
I think you are right in regards to A and B.
As for C, I don't have an answer. I will think about it.
Applications for a chromatic shift parameter:
Non diatonic intervalic doublings.
Pushing an entire line noticeably out of tune. Think of the classic Loony Tunes scores -- Carl Stalling would often take very recognizable upbeat classical quotations and push the instrumentation chromatically around to give them an absurd broken comedic effect (I compose music for Cartoons... thus the interest there)
Literal distortions -- say I draw a ramp up in the chromatic shift, either on an instrument or in a container to effect all instruments (or a portion of them). Sure I could do this by altering the progression, but the difference in the time it would take to do that is really extreme. If i need the progression for the accompaniment I can simply reharmonize, perhaps into a new layer.
In general, a chromatic shift parameter is more appropriate for longer passages and very obvious changes. It is musical but outside of pre-20th century practices to some degree.
Its main advantage is that it can be seen and thought of as a separate conceptual parameter that can be composed structurally outside of the chordal structure.
It is a clear display of the deviation from the norm that does not follow the notes as they move.
Ideally, eventually synfire would have a set of functions baking the shifts back and forth into different parameters.
What I mean by this is, the chromatic shift could be baked into the chord progression as altered extensions, as a separate harmonic layer, by reharmonizing, or into whatever figure based chromaticism ends up being implemented.
Like wise chord extension could be baked back in as figure based and/or parametric chromatic shifts to give us a basic triadic harmony. For example V7 bakes to V, then change the harmony to I. You now have a I7 which you can either reharmonize or not. Change the harmony to a ii instead and you have a ii7 not a II7. Obviously this is a very simple example.
EDIT on that last one, you actually wouldn't necessarily get a ii7 etc but you would have symbols with concrete intervalic shifts effecting the harmony, in a proportion similar to the proportion of triadic tones to extensions of the original.
So say the original V7 was a V Maj7 instead; The music happens to be all triadic tones with the exception of a 7 in the bass guitar. That bass note would be the note with the chromatic alteration though not necessarily a 7th in a new harmony.
Relative to any harmonic context or symbol type the bass guitar note always retains its deviation from the norm.
Sun, 2009-08-30 - 00:53 Permalink
just a remark
A) Chromatic chord movements
B) Chromatic melody alterations (blue notes, walking bass)
C) Precise definition of intervals between independent segments, possibly even across different instruments (which is what I understood Tim was asking for)
For me (A) clearly indicates a change in harmony, i.e. the desired chords have to be worked into the progression. There is no limit as to how wild or quick these can be. This also has the advantage that all instruments nicely respect the changes and, as long as the global key is retained, find their way across that "bridge" in a natural fashion, melody-wise.
In regard of Point A, this seems to me to be a very "harmonic" view. There are in jazz lots of chromatic chord movement without changing the general harmonie, just like chromatic in melodies.
So maybee bass and piano left hand stay unchanged, the right hand is playing outside or passing chords.
Or simply the playing of a chord half step above or below ending in the "real" chord.
Case (B) might be more easily expressed through the figure as explained above. A walking bass segment might have the anchor at a certain chord interval and move to/from that in chromatic steps.
Yes. (I guess that You mean by "chord intervall" a certain skalestep, part of the harmonie tones 1,3,5,7.) But, it must consider the surounding notes also.
Whenever the import recognizes more than one halfstep directly following, it can keep this, locking for a good endtone. And this could be an option, which can be used or not.
The following of the quarter notes of a walk bass over
Progression F7 / BB7:
F, A, C, Cb / Bb ...
Now lets say, the F7 chord stays for more than one bar.
Then the first 3 quarternotes F, A, C, are the same as before, but the best ending tone in the beginning of bar two seems to be an A, half tone above is Bb.
so the line with two bars of F7 chord would be
F, A, C, Bb / A.
The chromatic dissappeared, but the logic of the line is followed.
Playing F, A, Cb, Bb / A would give a terrible sound.
Possible would be F, A, D, Db / C.
So chromatic passing tones need to consider the surounding Notes.
---
Probably what I wrote is obvious to You.
Anyway, Chromatic is so common in that music, that the lacking of it feels strange.
[quote]
The most difficult case is (C) when used across different instruments, because there are so many other parameters that influence the final result: Playing ranges, interpretation settings, etc.
The simple "Chromatic" parameter is no problem, technically (basically it's just midi postprocessing like Shift). I would however prefer to know whether there are at least a few more applications beyond intervallic doubling. Any ideas?
When a composition is ready, the line in question could be rendered to static pitch, and can be duplicated. But of course this is not very comfortable for checking out ideas.
Sun, 2009-08-30 - 01:04 Permalink
The image you posted me reminds me of another aspect I have been thinking about which might solve that and be useful in other applications.
I know these are a lot of suggestions but maybe it makes where Im coming from clearer.
What if it were possible to tag groups of notes - like a segment but a note could exist in multiple tags.
The gui on this would be very simple.
This would:
Give Synfire the ability to see the multiple compositional relationships of the notes on multiple time scales, to better understand the context and voice leading.
Allow synfire to alias arbitrary groupings of notes. This would make the doubling issue far easier. The alias could be graphically shifted in all directions on any instrument. Changes to the original would effect all occurrences
Avoid confusing the musical function of segments with the needs of other features, say articulations.
A way to organize the composition with readable comments on the tags. As well these could be useful to give context to the phrases when the are being pulled from the phrase pool
Create selection sets (these could optionally be defined as musically inactive -- not considered in the same way that segments would be by Synfires engine). Say to repeatedly grab all the bass drums and change their type or increase their velocity over the course of a composition.
Theres mores... but for that image...
I could imagine having the segment grouping shown but the top notes "muted" (so as not to play on that instrument) tagged, and aliased somewhere else with those locked intervals.
Sun, 2009-08-30 - 01:20 Permalink
When a composition is ready, the line in question could be rendered to static pitch, and can be duplicated. But of course this is not very comfortable for checking out ideas.
Yes thats the main issue. We can always use static pitch.
I think it is essential to be able to work around these shifts while composing, especially in an orchestral context where the sheer number of layers that could potentially clash in a distasteful way is so much greater.
A walking bass is chromatic in passing moments. The dissonances Im concerned with often extend for long durations of time.
More so than in Jazz, where chromaticism is just the language, this chromatic shift parameter is set to conjure an obvious deviation from the language for effect.
The harmony and the deviation are easily separated which is why a simple parameter would serve it so well.
Sun, 2009-08-30 - 11:04 Permalink
. . .
I think it is essential to be able to work around these shifts while composing, especially in an orchestral context where the sheer number of layers that could potentially clash in a distasteful way is so much greater.A walking bass is chromatic in passing moments. The dissonances Im concerned with often extend for long durations of time.
More so than in Jazz, where chromaticism is just the language, this chromatic shift parameter is set to conjure an obvious deviation from the language for effect.
The harmony and the deviation are easily separated which is why a simple parameter would serve it so well.
Would it be helpful, to just add a chromatic option on parameter transpose? So segments, phrases up to longer parts of the compositzion could be transformed into any interval and add tension as wanted.
Or passing chords or tones in melodies could be manually done.
Users could be warned that this option can add very much tension, and should be used only by experiecnd musicans.:-D
(To be open, I´didnt read the whole thraed too carefully, so maybe I just repeated arguments.)
And it would be a first step into the chromatic world, which is now lacking in SF completly, as long as harmonie does not change.
So why not add this option.
(A manuel possibility for chromatic is for me personally not so important in the moment, because I prefer global transformation of phrases absoutly, doing long playbacks for jazzstandarts with dozens of chorusses).
So I´m much more interested in integration of a chromatic language.
As you wrote, a walking bass is chromatic in passing moments and its part of the jazz language.
So it should be added optionally by SF, even on places, where chromatic passing is not given in the orginal midi material, but maybe in other places leaving out given chromatic, because a diatonic solution is better.
(So in a progression of F7 / BB7:
F, A, C, Cb / Bb ... turns into F, A, C, Bb / A ,
when there are two bars of F7 chord.)
The chromatic dissappeared, but the logic of the line is followed.)
This is of course a more complex task.
Alfred
Sun, 2009-08-30 - 11:47 Permalink
I thought about a chromatic option in transpose, it would be convenient, but that parameter is considered by the harmonic server.
Lets say I copied a phrase of a flute, that had a transposition parameter attached to it, and doubled it on a clarinet; the clarinet would have to go through the same transformations as the flute, including the transpose, rendered to a midi note, based on the harmonic context and then shifted as a final step.
Again this might be problematic, as I don't know if Synfire is trained to make the clarinet and flute avoid each other in the same octave or not. The shift parameter is useful either way.
I also want to see chromatic options related to jazz and all kinds of music really. I think ideally the extended harmony would be able to interchange with the chromatic alterations to quickly adjust it from either perspective -- melodic or harmonic. Personally I'm much more comfortable dealing with them melodically and letting synfire pick up the correct harmonic alterations in the harmonies.
How would you go about adding chromatic notes to a bassline? As it is I don't think Synire ever generates its own rhythm (am i wrong on this?).
I think down the road a "Tonal" parameter would be cool to adjust, to a varying degree, how much chromatic content and non triadic content is let through the flood gate in a targeted, region specific way.
Of course there is no jazz chromaticism or classical chromaticism but having the attitude and approaches of varied styles and interests can give synfire a much broader range, if somewhat intangible.
Many ways to skin a cat is a phrase that is often muttered in the geeky habitats of software dev... Not only can you get Function X in the menu, but right click... There you go, presto!
I think its really amazing, that with this tool we are able to reapproach the fundamental components of music, separated from the whole, from so many different conceptual angles and visual correspondences.
Thanks to you I'm going to end up with english horns who think they're walking basses :)
Sun, 2009-08-30 - 15:46 Permalink
I thought about a chromatic option in transpose, it would be convenient, but that parameter is considered by the harmonic server.
Lets say I copied a phrase of a flute, that had a transposition parameter attached to it, and doubled it on a clarinet; the clarinet would have to go through the same transformations as the flute, including the transpose, rendered to a midi note, based on the harmonic context and then shifted as a final step.
well, this transpose option must have the speciality to simply transpose in absolute steps what would be the result
of the rendering of the orginal. Dont know how easy this would be in the software programming.
But anyway, i would be a first step. It would exclude drawing of chromatic or recognizing it in the import.
Andre gave some hints, that he has already ways for allowing that.
Again this might be problematic, as I don't know if Synfire is trained to make the clarinet and flute avoid each other in the same octave or not. The shift parameter is useful either way.
Why shift? Maybe you mean transpose instead?!
I also want to see chromatic options related to jazz and all kinds of music really. I think ideally the extended harmony would be able to interchange with the chromatic alterations to quickly adjust it from either perspective -- melodic or harmonic. Personally I'm much more comfortable dealing with them melodically and letting synfire pick up the correct harmonic alterations in the harmonies.
For me its mostly different. I have the harmony progression ready, ad then look for filling it up with bass, chords and soli.
How would you go about adding chromatic notes to a bassline? As it is I don't think Synire ever generates its own rhythm (am i wrong on this?).
I would define strong and weak positions in the bar, which is by the way an interesting thing in itself. A chromatic passing tone of a quarter on counting the one of the bar can be absultely weak, if followed by a long tone (3/4 f.e.) on the counting two and sounds therfore very natural and "diatonic". Strong and weak positions depend also strongly on the surrounding rhytmic positions.
But for the beginning it would be ok, just to place passing tones on positions like offbeats or the last note before a chord change. Quite often the first tone of a new chord is approched by a half step below or above.
I think down the road a "Tonal" parameter would be cool to adjust, to a varying degree, how much chromatic content and non triadic content is let through the flood gate in a targeted, region specific way.
:D
[quote]
Of course there is no jazz chromaticism or classical chromaticism but having the attitude and approaches of varied styles and interests can give synfire a much broader range, if somewhat intangible.
Yes.
[quote]
"Many ways to skin a cat is a phrase that is often muttered in the geeky habitats of software dev... Not only can you get Function X in the menu, but right click... There you go, presto!"
Sorry, your english is very high, mine is not so far developped. I don´t understand your first sentence here.
best regards
Alfred
Sun, 2009-08-30 - 19:25 Permalink
Andre,
Thanks for your help. Yes, chord symbols works well for specific scale steps if the step is part of the harmony. Thank you. I can certainly live with that - it works exactly as I expect. However, when the symbols are set to arpeggio the scale steps do seem to change from one harmony to another even when "strategy" is set to none. Not worth worrying about that, I think.
The "constant interval" relationship is more complicated than I thought. So obviously a medium to long term goal.
Many thanks!
Mon, 2009-08-31 - 11:47 Permalink
[quote="Tim UK"]However, when the symbols are set to arpeggio the scale steps do seem to change from one harmony to another even when "strategy" is set to none.
Hi Tim, the arpeggio symbols map to the notes in the order of pitch as they occur in the current inversion of the chord. IMO they are of limited use at this time (except for electronic music), because one has little influence on the chord inversion. It is determined dynamically when the alignment switch is on.