The previous post provided a brief overview of the OP-1's FM synth engine.
As usual, I will now attempt to identify some of the possibly problematic aspects of this synth. Again, it is with the utmost appreciation for the hard work and genius efforts of Teenage Engineering that I offer these pieces of feedback; I merely want to address any and all of the rough edges that exist so that future versions of the software can be even more perfect and awesome.
Let's begin!
I should start by pointing out that it's important to appreciate that frequency modulation synthesis is inherently very complex and inscrutable, and that TE have done a commendable job of making it (relatively) fun and friendly.
The fundamental problem here is that the parameter space is utterly huge; it's very easy to get lost in this sea of virtually infinite variety, and very difficult to navigate this vast unpredictable space in a cogent and directed manner. Basically, it often feels like you're rolling the dice rather than making decisions.
I'm not sure if there is any silver bullet here -- it might just be an intractable problem and there's no good way to provide the power and depth of FM without also overwhelming the user to some degree.
Visual System
The graphical display of FM is fairly straightforward and simple: some mathematical shapes and some numbers.
Blue and Orange are well-served by this display -- they're both abstract 1-dimensional values which are visualized as 1-dimensional movement (brightness and x-position respectively). Perhaps it would be nice to have a less-vague idea of Blue's current value, but this isn't a deal-breaker -- maybe I should just loosen up a bit and not be such a control freak!
Green and White, in contrast, are really not well-served by the current visual design. They are both very complex controls, and would require a lot of information to be conveyed to the user for he or she to be able to masterfully adjust them.
Sadly, none of this information is provided; unless the user is a savant and has memorized the relative positions of all 9 topologies and 102 frequency sets, then they're at a loss as to how to get from their current state to any other desired state, and all they can do is roll the dice and randomly hunt around for the particular topology or frequencies they want. Knowing what you want but not knowing which way to turn the knob in order to achieve this goal state is a very frustrating and disempowering user experience.
In the specific case of Green, this can be a very tedious hunt as there are dozens of options to scroll through, and you have to be careful not to carelessly skip over your desired frequency set.
A better solution to the problem that this embarrassment of options poses might be to provide the user with some contextual information in order to allow them to guide their search more effectively.
Perhaps when White is turned, the view could zoom out to show all 9 different topologies -- the user could then very easily and naturally scroll the selection to the one that they desire. A different graphical treatment which provided room for all the topologies (and ideally made them less vague -- it's not really clear in some cases whether operators are modulating, mixing, or outputting!) to be present constantly on the screen (with the current option highlighted) might be an even better solution.
In the case of Green, it's likely not possible to display all the available options on-screen at once; at the very least showing some of the adjacent/neighbouring values would provide some context that would help the user locate their current position in the parameter space, as well as making it easier to quickly scan through the values without missing your particular target -- there will be some forewarning as it approaches in the list, and even if you don't stop directly on it, you'll see it nearby. Some way to aid in the navigation of the frequency sets beyond the current "absolutely no idea where I am or where I'm going" would be greatly appreciated!
Parameters
As with the visual display, the main issue with FM's synthesis parameters is that Green is currently not as useable as it could be.
For starters, a simple solution would be to reduce the number of options to something a bit more manageable -- I appreciate that TE have given us a ton of creative space to play around with, but frankly many different parameter sets sound quite similar to me. Possibly the problem they faced was that some sets sounded similar with a particular topology, but distinct in a different topology -- in this case, why not make Green dependent on White, so that for each topology there is a manageable set of frequencies to choose from, rather than the current overload of options?
Since reducing the choices available the the user is seldom a great solution, what else could be done to make Green easier and more fun to use?
One relatively simple solution to empowering the user would be to lay out the different frequency sets in a more intuitive order. Currently they're simply sorted numerically, which means that the inharmonic sets (typically anything with awkward ratios like 7/16) are interspersed throughout the harmonic sets (which are typically -- though not always -- those with integer ratios).
If the frequencies were sorted perceptually rather than numerically -- with similar-sounding sets adjacent to each other -- this would go a long way towards both making it easier for the user to locate and navigate the space, and making it easier to tweak Green live without sudden unpleasant surprises. Such ordering is likely to be somewhat topology-dependent, but there is definitely some sort of sensible musical order than can be applied -- after all, 1-2-1 will always sound more similar to 2-2-2 than to 1/2-5/32-7/16.
A different solution to the problem of Green useability would be to note that the Orange control is currently not all that useful: the frequency sets already allow the generation of all manner of clangorous inharmonic tones, being able to make them even more so is less important than being able to control them a bit more.
My proposal is this: the frequency sets could be split into two, with the harmonic sets in one list and the inharmonic sets in another. These lists would be of equal length, so that each harmonic set is paired with a corresponding inharmonic set of similar frequency; for example 5/8-2-2 and 5/8-1.5-1. (Ideally, each list would also be arranged in a perceptual order so that when sweeping through them, the sound would slowly change rather than jumping all over the place) Then, Orange could be used to smoothly morph between the currently selected harmonic (Orange fully CCW) and inharmonic (Orange fully CW) pair. This would allow Orange to function both as a detune (when it's close to the CCW and CW limits) and as a "how inharmonic do I want this?" control.
Or perhaps it makes more sense for there to be three parallel lists, with two complementary harmonic sets bookending an inharmonic set?
Anyway, this is just an idea -- more important than such OTOH inventions is to embrace their underlying principle: that it would be better to approach the organization and arrangement of the frequency sets in some comprehensible musical way rather than just sorting them numerically and throwing them in our laps.
There must be some more useful, interesting and novel way to approach FM than the current setup, which I feel is definitely a good step in the right direction, but not all the way there yet in terms of making frequency modulation synthesis useable and fun.
On a related note: currently, tweaking either Green or White while a sequence is playing live is akin to Russian Roulette. Even if you have identified and located some suitable frequency sets or topologies ahead of time and know their relative positions, you still need to quickly navigate between them during pauses between notes. This becomes even more difficult in the case of White, as here the "parameter mapping" between encoder and internal value is loose and sloppy -- it takes a variable number of encoder steps to move to the previous/next topology -- with the result that it's difficult to pull off fast precise movements.
Finally, the insanely rich nature of the timbres which FM is capable of generating makes me really wish there was more I could do with it: a multimode filter, an enveloped filter... basically anything that would let me shape and sculpt the dense sounds would be greatly appreciated.
In Summary
OP-1's FM synth engine is definitely powerful and useful -- it can handle anything from deep thick bass to soaring sweeping pads to searing screeching leads, and plenty of clangy bell and warbling insane-demon-banshee as well.
Currently this power feels a bit "raw" and unsafe -- and perhaps this is intentional, or an inescapable part of frequency modulation. But I do think that at least some things could be tweaked or re-assessed so that the user is more empowered and in control rather than randomly wandering through a vast and incomprehensible parameter space.
Anyway, that concludes our look at OP-1:FM. If you have any feedback, comments, criticisms, corrections, suggestions, complaints, revisions, additions, thoughts, or you're just bored and want to talk about FM, please hit this forum thread and let the discussion begin! http://ohpeewon.com/discussion/231/op-101-fm
