String: Critique

The previous post attempted to cover the basics of String, describing its inner and outer workings, and attempting to unravel its mysteries. (That was a pun: string, unravel... okay, it was weak, I admit!)

Now our focus shifts towards the more problematic aspects of the interface and design of String.



Visual System
String's graphics are among the most colourful and pleasant of any of OP-1's synth engines; they are useful and functional and perfectly okay... but not perfect.

The most striking aspect of their design is the juxtaposition of imaginative informational graphics (the White microphone placement to indicate phase offset is a wonderfully succinct and expressive idea) with straight-up bar graphs (the Green and Orange graphics). There isn't even really an attempt to unify the two linear graphs visually, each having a distinctive look that's starkly utilitarian compared to the animated string graphics.

This seems either lazily or hastily designed, sort of a thrown-together grab bag of ideas rather than a unified cohesive design (such as that demostrated by the elegant and beautiful hardware).

Of course, the line graphs are perfectly serviceable and do provide excellent visual feedback about the parameter in numerical terms: current, minimum, and maximum values are clear and immediately apparent, and this shouldn't be scoffed at since many of the other synths are not so thoughtful and clear in terms of visual communication -- even String's own Blue parameter is visualized in a very vague and awkward way, so maybe sticking to line graphs isn't such a bad decision.

However, I can't help but feel that the graphical display of the synth is being wasted here, and that some attempt to better communicate the meaning of the Green and Orange parameters might be fruitful: perhaps by showing a pair of ramps (corresponding to the impulse's decay and spectrum/filter settings, which are controlled by Green and Orange respectively) or some other equally clear but more informational visualization.

Again, as with so many of the synth engines' graphics, the ability to communicate meaning in addition to strictly numerical state is latent in the design but lost in the execution, failing to live up to the strikingly beautiful and confident modern design ethos which is masterfully evinced by the hardware itself.


One final criticism of the graphical element, which also pertains to the choice of parameter mappings themselves, is that the visual on-screen layout and physical ordering of encoders seem to be at odds, and strangely confused.

For one thing, just making the visuals consistent with the encoders would be nice; currently Green is the leftmost parameter on the display, and -- coupled with the bizarrely perverse use of blue rather than green for the min/max labels -- this can cause some confusion, as one naturally reaches for Blue rather than Green. This could easily be remedied by swapping Blue and Green encoders.

However, this would be a kludge rather than a proper design solution; the underlying dissonance between visual, physical, and functional layout would remain unresolved.

The actual signal path of this synth, if I'm not mistaken, is something along the lines of: Impulse -> Filter -> Phase Offset.

This suggests that the most logical and intuitive parameter layout would be for Blue and Green to control the two impulse controls (Impulse Decay and Impulse Type), White to control Tension, and Orange to control Detune. This way, left-to-right movement across encoders corresponds to movement through the signal path from start to end.

This more rational layout would have to be accompanied by a corresponding visual treatment; if the string and microphone graphics (the current visualization for Blue and White) were moved to the right side of the screen, this would free up room on the left side for some sort of animated info-graphic display to communicate the impulse's properties to the user. Perhaps a waveform would be useful; I'm sure TE would have better ideas for how to communicate the impulse's properties than I might be able to offer.


Parameters: Blue
This is the most visually ambiguous and vague of String's parameters, and unfortunately also the most dangerous, as too high of a setting will cause the resulting notes to sound flat and out of tune. The obvious solutions to this perfect storm situation would be to find some clearer and more precise way to communicate the current state of Blue, and/or to remove the sections of the parameter space where undesirable behaviours are found.

Obviously, what one person deems undesirable another might consider a useful tool, but in this case I hope that agreement is not difficult: the way that the pitch goes out of tune at high Tension values is awkward and annoying. Either limiting the parameter to avoid this problem zone, or compensating for it in some way (by artificially increasing the pitch to correct it back into tune) would be good steps.

Visually, the isometric string array is very pleasant; having the strings respond visually as notes are triggered is a nice touch, and helps add a secondary path of visual feedback to the parameters, but this is insufficient as the sole visual feedback, as is the case with Blue.

One idea to unify the visual design would be to locate a bar graph along the bottom left edge of the base of the isometric graphic -- i.e perpendicular to the current Green bar graph, which is located along the back left edge. (The Green graphic could also stand to be toned down to the minimal simplicity of the Orange bar graph) White could similarly be represented as a white line extending across the side currently spanned by the microphones; in this way, the animated string array would be framed by the set of parameters represented as simple coloured line segments.

Anyway, this is just an off-the-cuff idea; ideally all the graphics could be illustrated in the same charming and useful way as White currently is (i.e with pictographic/diagrammatic animations), but otherwise it would be nice to unify them into the simplest possible presentation: bar graphs.

One final comment on Blue is that its function is somewhat confusing, being backwards/opposite to the typical synth knob behaviour of "decreasing/shorter decay as I turn CCW".

Because Blue is labelled Tension, it makes a certain kind of sense to have increasing tension (CW) lead to shorter and faster decay. However, in this case, an increasing abstract parameter leads to an observed decrease... for all intents and purposes, Blue is a decay control.

This results in that most heinous of UI flaws, the dreaded negative transfer, as the control mapping works opposite to that of other controls in the OP-1 (and indeed most every other synth in existence).

It seems like reversing the behaviour of this control (so that CCW = increased tension = faster decay) might make more sense; if it pains TE to present decreasing tension as a CW movement, they could simply relabel this control "Decay", or even "Filter" (similar to other OP-1 synth engines), which would be both accurate and descriptive, while allowing a more intuitive control mapping: faster = shorter = less = CCW.


Parameters: Green
The Green parameter currently seems to control a downward-ramp envelope, applied to the impulse's volume. This is a bit boring for a few reasons, primarily because the synth already has a decay-shaped control (Blue), and since that parameter is inherent to the very nature of Karplus-Strong synthesis (time's arrow and all that), it would be better to allow the other parameters to provide some sort of contrary dynamic, pushing out to expand the parameter space and palette of the synth.

I'm not entirely certain what would happen were the impulse to be shaped by e.g a triangular envelope with adjustable attack, but I suspect this would nicely emulate bowing of the strings -- the impulse being gradually strengthened, analogous to some bowing techniques. Were a sine or triangle wave used to modulate the impulse's level, this might effectively mimic the "vibrato" bowing technique where the musician is bowing back and forth with very quick and short movements, injecting a rapid succession of tiny impulses of vibrational energy.

Given that many interesting envelope controls exist elsewhere in the OP-1, if my above assumption is correct and the useful range of envelope shapes for the impulse's volume extend beyond merely downward ramps of varying slope, it might be useful to introduce some sort of variable triangle envelope where the user could control both attack and decay rates.

One idea for such a control would be a "bipolar" parameter where center and maximum would be mapped to the current minimum and maximum settings, i.e turning CW from center would be identical to currently turning CW from minimum: the decay time would steadily increase. On the other half of the parameter space, turning CCW from center would gradually increase both attack and decay time, creating an isosceles-triangular shape of variable slope. Such a control scheme would make it impossible to achieve a purely upward-ramp shape, but perhaps such a shape isn't as useful in this context.

(Or, just make the parameter control impulse level and give us some more envelopes and LFOs to modulate parameters with... but I'm afraid that this is sadly too much wishful thinking.)

Of course, it may be hubris and folly for me to suggest such ideas, since TE have no doubt spent a lot of time tinkering and brainstorming and iterating and experimenting; perhaps they have already tried many approaches and the current design is objectively the best possible control scheme.


Parameters: White
This parameter seems to be visualized backwards: as the phase offset increases, the space between the microphones decreases. Surely this is a bug or an error, since it completely ruins and undermines the wonderful microphone metaphor (i.e tapping the vibrating string at two different points along its length to get two out-of-phase waves); when the phase offset is zero, the microphones (or whatever those two little white things are) should be coincident, moving apart from each other as the phase offset increases.

This actually reminds me of the similar seemingly opposite visualization of Pulse's Orange parameter, where AM and PWM animations appeared to be swapped for each other. Part of me can't help but wonder whether this is honest error or oversight on the part of TE, or whether it's an intentional conspiracy of theirs -- an insidious plot to undermine our faith in our eyes and thus forcing us to rely more on our ears when making music.

This is certainly a noble and notable goal, but perhaps reversing visual and aural feedback -- creating a lot of confusion and cognitive dissonance in the process -- is not the best approach.


Parameters: Orange
I don't think there's anything to say here that wasn't covered above, i.e since it's adjusting an aspect of the impulse -- Impulse Type -- it would make more sense to locate it near Impulse Decay, and it would be nice to have a visual which communicated more the meaning of the parameter (e.g a spectrum or cutoff/slope diagram) rather than its abstract mathematical state.


Summary
As it stands, String is a very pleasant and useful component of OP-1's range of synthesis options.

While it certainly isn't as diverse or dynamic as some of the other synth engines -- being inherently limited to sounds with a decaying timbre -- it still has a decent amount of range. On the whole it does tend to, perhaps understandably, always sound like a plucked string, but more extreme parameter settings can sometimes coax it grudgingly outside of this staid zone and into more interesting sonic territory.

Perhaps its most obvious flaw is the way in which it seems reluctant to gel nicely with the other synth engines; its distinct sound is strikingly different than the other, more digital/artificial-sounding synths, which means that listeners are immediately aware of its presence in a song -- "oh, there's String!" -- and this can be somewhat awkward.

Obviously, this is one of the many aural limitations which are to a certain extent an unavoidable consequence of the Karplus-Strong synthesis method; for instance, tweaking Green or Orange during the sustained portion of a note produces no audible effect whatsoever, and it's hard to see a way around these sort of inherent constraints.

Nevertheless, I do think that further control over the impulse shape -- both its volume envelope and filter/colour -- might allow a broader range of sounds to be generated, and I hope that at some point TE turn their ingenious minds towards souping up String a bit and giving it more room to move around in.

If you have any comments, criticism, suggestions, corrections, or any other feedback about this article or String, you can discuss it here on the forums: http://ohpeewon.com/discussion/239/op-101-string


Afterword
This concludes the first series of OP-101 articles, which have covered the current bevy of synth engines. There may in the future be other posts, concerning the many other interesting and intriguing aspects of this marvelous device, but for now I'm going to take a break.

If anyone wants to contribute a guest-post or has any suggestions or feedback, please just post or PM me on the forums -- it would be great to hear from you!

Thanks for reading :)