The previous post discussed the basics of the Pulse synth engine: an oscillator composed -- unsurprisingly -- of two pulses, whose amplitude and relative position could be manually adjusted as well as modulated by a dedicated LFO, before being fed through an adjustable lowpass filter with a decaying cutoff envelope. At least, that's my best guess! :)
As always, I'm now going to try to discuss the most notable flaws or awkward design decisions of this synth engine, in terms of both synthesis and user interface. It bears repeating that this isn't intended to be a criticism of Teenage Engineering or an indictment of Pulse; it's simply useful and interesting to identify, examine, and consider the weakest aspects of a design -- especially when the design is software-based and thus relatively fluid and malleable.
Let the scrutiny begin!
Visual System
The graphical presentation of Pulse is quite busy, but this relative abundance of visual feedback is actually very useful and useable -- at least, it could be.
The integration of a waveform display (the blue double-rectangle waveshape in the top center of the screen) is undermined by its inaccuracy; in some cases it seems like it's doing its best to lie outright to users, as when CCW values of Orange produce variable-amplitude shapes while CW values produce variable pulsewidths -- the exact opposite of the actual audible behaviour!
Similarly, the response of this waveform to Blue, Green, and White controls lack any real correspondence to the audible effect of those controls. At best they are impressionistic, but they are always quite inaccurate and as a result of limited practical value.
In the case of Blue, the waveform visibly smooths into a sine shape as the parameter is turned CCW -- assumedly displaying the result of lowering the filter's cutoff. However, audibly this adjustment can have the exact opposite effect: when playing notes in low octaves, lowering the filter's cutoff causes huge amounts of overdriving and clipping -- either through increased resonance or some other waveform-distorting behaviour. This means that, as with the aforementioned Orange visualization, the visual and aural feedback experienced by the user is contradictory... results which are at odds with each other.
Turning to the lower portion of the display, the dual pulses represented there are similarly disingenuous: Green appears to alter the level of the first pulse, while in actuality it controls both pulses. White on the other hand appears to move the second pulse between 20% and 80% of the way through the oscillator's period; in reality, the second pulse is moved from 0% to 50%.
This lack of correct correspondence creates a mental dissonance when trying to comprehend cause and effect: for example, you would imagine that visually locating the second pulse directly between the two pictured primary pulses would result in a doubling of the perceived pitch, as such positioning would effectively halve the oscillator's wavelength. Instead, moving the second pulse halfway visually results in nothing close to this effect, while moving it to the extreme right visually, at the very end of the (visual) period does result in the previously anticipated pitch change.
As with some other corners of the OP-1 (e.g Delay's Orange/Input parameter), Pulse's visual interface keeps trying to hide useful state from the user: after a short period of inactivity, the Green and White labels ("H" and "T" respectively) both fade into the background. This is perhaps an attempt to de-clutter the lower graphic, but in practice what this results in is hiding the parameter mapping from the user: it's no longer immediately obvious and clear which encoder should be turned to affect a specific desired change to the displayed pulses.
This is frustrating, and also difficult to justify: is there a real benefit to this behaviour? It certainly seems like it took more programming effort than simply redesigning the graphic in order to avoid such awkward corner cases.
Moreover, it clashes with the behaviour of the other two parameters: the "F" and "M" labels (for Blue/Filter and Orange/Modulation) don't fade out after inactivity.
The way in which the H and T labels tend to shift and move around to get out of the way of the pulses seems decidedly awkward and kludgey: locating these labels in the middle of the line segments (the linear intervals demarcating the top/bottom of the first pulse and the gap between the first and second pulse) creates unnecessary problems which awkward special-case movement then attempts to correct.
A better solution might be to locate the H label to the left of its associated interval graphic, and to move the T interval and label to below the pulse shapes rather than between them (drawing the interval directly below the pulses while the label itself is below the interval rather than bisecting it). Hopefully this explanation is clear.
Finally, it's quite frustrating that while some sort of permanent hard-coded decaying modulation is clearly audible, the state of this modulation -- its shape, rate, amount, etc. -- is nowhere displayed. Keeping this behaviour hidden but heard is another source of cognitive dissonance between visual and audible feedback.
A visual display which appears to the user to be an informative literal depiction -- i.e a waveform -- but which is in effect an artistic reinterpretation rather than scientific rendering, is at best misleading and at worst counter-productive, as it can inhibit the user's ability to form a useful mental model if it suggests relationships which are not consistant with perceived reality. In other words, if the audio and the visual feedback don't match, the interface tends to become a hindrance rather than a help.
The visual system of Pulse is frustrating for this reason, because the graphics come very close to being transparent, but then veer off-course and instead obfuscate and mislead.
Certainly it is quite effective at communicating the current value -- as well as the maximum and minimum values -- of the parameters, but this is merely the most base and standard application of visual feedback. The beautiful display and vector graphics have the capacity to communicate much more to the user: to inform them of the meaning or implications of each parameter. Sadly in this case such capabilities aren't fully leveraged.
Parameters: Blue
This is possibly my least-favorite parameter out of any synth in the OP-1. Perhaps not coincidentally, it's also one of the parameters most likely to ruin everything when tweaking live.
I'm not sure whether it's the filter itself, or the set of variables controlled by this parameter, or simply the chosen range of these variables, but this is one of the least-musical, worst-sounding filters I've heard.
Its sound jumps from weak, subtle and understated (Blue close to maximum) to horribly overdriven clipped out-of-tune mess (Blue close to minimum) with an awkward unuseable range in the middle where the signal is softened/filtered so much that any set of Green/White/Orange values tend to sound the same -- a soft sine-ish tone without any useful character.
Rather than shaping or sculpting the sound of the oscillator, this filter sounds like a "tone" control on a cheap CD player for the upper half of its range, and then bewilderingly blows out the oscillator in the lower half of its range.
The clipping in the low-end is definitely not nice-sounding, or even appealingly raunchy/evil/ugly: it's just purely out of control, and it ruins the perceived pitch of the synth, muddling things into a terrible mess that is really not useful or useable in the vast majority of cases.
Additionally, the way the cutoff seems to only half-track the keyboard means that even if you do manage to tweak Blue into a decent sound, this will only work for a few adjacent notes: jump up an octave and the signal getting weak and washed-out (overly filtered), while jumping down an octave will lands you squarely in mega-distorto land.
I think that an overdriven filter is a great idea and would work well with many of the OP-1's synth engines, but in this case it doesn't sound like an intentional -- i.e musical and pleasing, controllable -- effect but rather like a DSP error.
Basically, this parameter sounds good and is useful10% of the time, and the rest of the time it's either somewhat useless (functioning as a combination EQ and volume control, when there are multiple superior versions of such controls in other areas of the OP-1) or actively counter-productive (completely filtering out high notes or completely ruining low notes into a blown-out mess).
Blue/Filter actually seems buggy or broken; it stands out markedly from the rest of the synths and parameters in the OP-1.
Parameters: Green
The Green parameter is relatively staid: a glorified volume control which may additionally change the timbre in very subtle ways. Any change in tone is completely drowned out by the much more obvious change in volume -- and given that there are many other ways to change and shape the volume of a synth engine's output, this seems like a wasted parameter.
Parameters: White
I don't actually have any criticisms to offer concerning White, aside from the disparity between its visual and actual behaviour as mentioned previously. It's a fun and novel control, with a relatively large useful range of tonal adjustments possible.
Parameters: Orange
While the modulations offered by Orange are quite useful and broad, they also seem somewhat redundant -- in the case of negative/CCW values at least, we could use the OP-1's LFO section to achieve the same PWM-type results without tying up a precious parameter/encoder.
Positive/CW values are sadly much less useful than CCW, as they are quite heavy-handed; certainly heavy AM is an interesting synthesis option, but the relative lack of precise control here (it's more or less impossible to tune the LFO's speed finely enough to ensure that the overtones/side-bands generated by fast modulation work with the notes you're playing) coupled with the overwhelming strength of modulation has the result that only the first couple of "notches" are useful, and those may even be too subtle.
Essentially, this control (in its positive range) shares the same basic problem as Blue: it jumps from "not enough" to "too much" without passing through a broadly useful range in between. As it stands, both are novelty parameters -- of limited use in specific narrow applications, but far from generally applicable in the majority of cases, being either far too subtle/boring or too powerful and uncontrollable to work with most material.
It should be mentioned that the bipolar style of control found in Orange is a very excellent idea, making the most of the OP-1's minimal set of controls. It would be nice to see more parameters which offer similar dual-flavoured behaviour throughout the synth.
In Conclusion
Pulse is a very interesting synth, and brimming with potential, but as it currently stands this potential seems to be mostly unrealized; it's just too inflexible and awkward in its current state.
The relative uselessness of most of the controls certainly contributes to the stiff, static, and constrainednature of this synth engine; Blue and Green mostly function as volume controls, whose timbral adjustments are either too subtle or too strong without a useful range in-between. Most parameters seem like a waste of that most precious of OP-1 commodities, an encoder, with the result that Pulse is the most undynamic synth engine in the OP-1 (barring perhaps the synth-sampler, but that's a matter for another day).
The feature which seems to undermine Pulse's wonderful premise the most is the inescapable pervasive downward-ramping decay: this ever-present modulation, whether produced by filter movement or some other process, is unavoidable, and inexorably pushes all timbres towards the relatively narrow range of "plucked" sounds.
This decision is doubly awkward when you consider that this is the same precise range of sounds which String occupies -- and since in this case the decay-shaped nature of the synthesized sound is inherent to the synthesis process (but more on that later).
If only TE could provide a greater ability to shape and control Pulse, it would immediately become much more friendly and useable.
I realize that they may not want to repeat themselves, but the enveloped controls of Cluster's Green parameter or the Random and Tremolo LFOs are quite powerful, expressive, and useful -- it would be great to see this kind of parameter exposed to help diversify Pulse's range.
The terribly uncooperative nature of the filter is also surprising, as many other synths feature quite pleasant and useful filter sections. I hope that this is the result of some sort of error or oversight, rather than an intentional decision, because it's quite starkly not friendly or fun, or even useable, compared to any of the other synth parameters available in all of the OP-1's considerable arsenal.
I really do like the uniqueness and novel left-field nature of the idea behind Pulse: it's interesting, exciting, and powerful! Sadly in its current form it just doesn't translate into a very useful tool. It's definitely my least-favorite of all the OP-1's synth engines.
If you disagree -- if you think I'm missing something or have made a grave error in judgement -- or if you have any other comments, criticisms, suggestions, complains, or just want to discuss Pulse further, please visit this thread in the forums and let's chat: http://ohpeewon.com/discussion/238/op-101-pulse
