Cherry Audio recreate Moog’s classic polysynth at a fraction of the price of an original.
Sometimes it seems that every vintage keyboard of repute has been copied, either in hardware or as a soft synth that promises to sound indistinguishable from the original. But there’s one venerated instrument that has escaped the attentions of the mainstream plagiarists, at least until now. The synth in question is the Memorymoog, perhaps the last of the great polysynths of the first golden age of analogue synthesis.
Despite its impressive looks, the Memorymoog is a fairly basic polysynth. It was inspired in part by the Minimoog, so each of its six voices has three oscillators, the first two of which can be sync’ed, and the third of which can act as a modulator at low and audio frequencies. The outputs from these are mixed (together with noise if wanted) and passed to a 24dB/oct resonant low‑pass filter and then to a VCA, each of which has its own ADS/AR/ADSR contour generator. Additional modulation is provided by a single, global LFO. Various polyphonic and monophonic modes are provided, the latter with unison so that, as well as being a superb source of polyphonic brass patches, pads and unusual percussive sounds, you can use the Memorymoog as a monster monosynth.
Its spiritual descendent, the Memorymode, is supplied in VST, AAX, AU and standalone formats, emulates the look and feel of its inspiration, and incorporates a number of enhancements that are designed to add flexibility without detracting from the original intent. Consequently, its oscillator and mixer sections are almost precise recreations, which isn’t as trivial as it sounds. Memorymoogs sound great provided that you follow one simple rule: unless you want the synth to become almost unmixable, don’t turn its oscillator levels to 10. In fact, don’t even turn them up to 5; some of the nicest Memorymoog sounds are obtained with the levels in the range 1 to 3 so that the signal chain isn’t overdriven. The Memorymode emulates this behaviour well. Turn the oscillator levels up and it sounds huge; turn them down and it can sound delicate, smooth or creamy as you choose.
The first obvious differences are revealed in the filter section, where the Memorymoog’s four filter tracking options (off, 1/3, 2/3, 1) have been replaced by a knob offering a smooth tracking range from 0 to 100 percent. Much more significant are the introduction of two new facilities: velocity sensitivity (which affects the contour amount rather than the cutoff frequency itself), and a new 12dB/oct option. The difference in character obtained when switching between filter modes is far from subtle, to the extent that you can almost treat the Memorymode in 12dB/oct mode as a different synthesizer.
Velocity sensitivity has also been added to the contour amount applied to the VCA gain, but both the VCF and VCA contours retain the unusual Return To Zero, Unconditional Contour, Keyboard Follow (higher notes have shorter attack, decay and release times) and Release On/Off options of the original.
You’ll also find changes in the Outputs section, which now includes an optional limiter and a Modern button that applies a mild smiley‑face EQ to the sound. However, the loss of the Memorymoog’s Programmable Volume knob isn’t a shortcoming because the value of the Memorymode’s Master Volume is saved as part of each patch. Below this, the Tune section offers a misnamed oscillator Drift control that, rather than allowing individual voices to drift, determines the maximum amount of static detune between voices.
The Memorymoog’s modulation options are similar to those of a Prophet 5, with a polyphonic Voice Modulation section plus a single, global LFO that affects all of the voices simultaneously. No doubt in the face of significant temptation, the Memorymode has preserved this architecture, albeit with enhancements. For example, the five waveforms and the rate control for the LFO are retained but there are now eight destinations, with the VGA Gain added to the previous options. The modulation initial mount control has migrated to this panel (which is sensible), a mod wheel amount control has appeared (which is also sensible), and three new buttons have been introduced. The first is Keyboard Reset which, when activated, reinitialises the LFO whenever you press a new key after all previous ones have been released. The second is Sync, which locks the LFO rate to MIDI clock and, in the standalone version, to the tempo defined in the menu bar. The third is DC which, when selected, causes the modulation wheel to generate the equivalent of a DC voltage and apply this to the selected destination(s). The way that this is applied to the VCA gain is unusual (and too arcane to explain here) but makes sense in use. The Voice Modulation panel also has an extra destination — again, this is the VCA Gain — but is otherwise the same as the Memorymoog’s.
Although limited by modern standards, these modulation options are more flexible than they might seem. But beware — the subtleties of analogue cross modulation are difficult to recreate in the digital realm. Happily, using Osc3 as an audio frequency modulator offers all manner of bright and jangly opportunities, (as well as AM via the new link to the VCA) but don’t expect Yamaha‑style 2‑op FM because you wouldn’t get that from a Memorymoog either.
For obvious reasons, the original synth’s System Controller and Foot Pedals panels as well as its Keyboard Out options have disappeared from the Memorymode’s GUI, but the before/after parameter value display has been retained, and the remaining facilities have been grouped into two new panels: Keyboard Mode and Arpeggio. A Memorymoog offers four polyphonic keyboard modes, which differ in how the notes played are assigned to the voice cards, and all four appear to have been accurately recreated. There are also six monophonic modes — last, low and high note priorities each with either single‑ and multi‑triggering, and you can determine the number of voices played in these modes using the System Controller. (A full 6‑voice unison stack is one of the reasons that the synth is revered to this day.) Again, the Memorymode emulates these but now with up to 16‑voice unison, and a detune knob is provided to thicken the sound still further. I find it amusing that I have spent many days over the years wielding miniature screwdrivers to ensure that my Memorymoog is in tune, whereas Cherry Audio has added a knob to ensure that it isn’t! The original also has a button misleadingly named Hold. This is a chord function, so whoever designed the Memorymode’s GUI sensibly renamed this Chord Mode. The next control is the Bend Depth, which is the same as the Pitch Bend Amount on the original and offers the same maximum range of ±1 octave. The final control in this section is the octave shift. The original offered just unshifted or ‑1 options, while the soft synth adds a +1 option to these.
The final stage in the Memorymode’s signal path is occupied by four digital effects that the manual describes as “super‑rad, studio‑quality effects”. Bah, humbug! They comprise a basic phaser, ensemble, echo and reverb, and they’re in a fixed series that’s in the wrong order. The classic combination of 1970s and early 1980s effects has the ensemble followed by an external phaser such as an MXR Phase 100 or an Electro Harmonix Small Stone. In the Memorymode, the phaser comes first. This offers a pleasing ’70s sweep, and a single PWM’d oscillator played through it is instant Oxygène, but I found the ensemble to be wholly uninspiring. The Echo offers delay times of up to two seconds as well as MIDI sync but, although it’s described as a stereo effect, this is accomplished by offsetting the delay time in the left and right channels as you increase the spread. With even a moderate amount of feedback, this makes the delay go out of time with itself. Finally, the reverb offers three Types — room, plate and hall. To be honest, you wouldn’t find me replacing any of my existing effects with any of these.
Programming the Memorymode is simpler than its wide and seemingly dense panel suggests, and it proved to be quick and easy to sculpt the sounds that I wanted. Having done so, I then used MIDI Learn to provide the performance capabilities that I wanted for each patch. (See 'MIDI Performance & Automation' box.) Once I had created my sonic masterpieces, saving and loading patches was much as I had expected, with factory categories and collections, user‑defined categories and collections, definable keywords and favourites.
Housekeeping is also straightforward. There are settings for things such as the GUI theme (classic or white), its zoom level (50 to 150 percent), how the mouse or track pad controls the knobs, and so on. On the standalone version, there are also settings for the MIDI inputs, the sample rate, the buffer size and the audio outputs. While there’s nothing radical here, it’s done nicely, and I must admit that I rather like the white GUI which — to my eyes — looks less dense than the classic panel (even though it isn’t). If there’s just one thing that I didn’t like about the system it’s that, if the host computer is online, the Memorymode takes advantage of this and searches for upgrades and new factory patches, downloading and installing these automatically if you allow it to do so. Cherry Audio also use this mechanism for in‑app advertisements. You can disable this, but I can’t say that I’m comfortable with a synthesizer acting in this fashion. It may be the way of the future, but I hope not.
Another concern is the lack of a printed (or printable) manual. What’s more, the online help attempts to be amusing throughout and, in my view, fails horribly. I accept that others might like the style but, if a manufacturer is going to pursue this approach, it had better be certain that everything else is in good order, and it isn’t. The descriptions include graphics copied from the company’s MS20 clone rather than the Memorymode, and there are several misleading statements to confuse the novice. It needs to be revisited.
Programming the Memorymode is simpler than its wide and seemingly dense panel suggests, and it proved to be quick and easy to sculpt the sounds that I wanted.
To test this, I placed my Memorymoog alongside the MacBook Pro running the Memorymode and did all of the usual things to ensure that the outputs from the two were matched. I then compared Cherry Audio’s recreations of the Memorymoog’s factory patches against the same sounds on the original. I found that the initial timbre is good, the filter retains its essential Mooginess and doesn’t lose its bottom end when moderate amounts of resonance are dialled in, and that the snappiness of the original had also been retained. For these and other reasons, some of the simpler factory patches sounded very similar to the Memorymoog’s, which is no small compliment. More complex patches — for example, those that use audio frequency modulation — were much less so, but I wasn’t surprised by this; when programmed in this way, small calibration differences can result in significant sonic differences.
Next, I performed a series of forensic comparisons, and the results of the filter tests were intriguing. The maximum sample rate of the soft synth is just 48kHz, which means that its bandwidth is a little over 20kHz. This was borne out when I tested the range of self‑oscillation of its filter, which was a creditable 20Hz – 20kHz, although this is narrower than that of my Memorymoog. But these tests revealed the biggest difference between the two synths. As you’re probably aware, it’s relatively simple to open a Memorymoog and correct the pitch and scaling of the oscillators such that the auto‑tune mechanism is then able to grab them and drag them into tune. However, it’s much harder to calibrate the filters. This means that, even when the filters are not oscillating, each voice in a Memorymoog tends to have a different character from the others. The consequence of this can be quite subtle on a well‑calibrated Memorymoog, but it adds hugely to the character of the instrument. On the soft synth, the filters are always perfectly in tune with one another, which removes the variation from the voices. More than anything else, this distances the Memorymode from the Memorymoog.
Talking of variations, I have to return to the Memorymode’s Drift control. If you select a handful of voices, apply a bit of drift, and then play the same note in Poly Mode 1 (voice cycling), you hear the dreaded da‑da‑da‑dee‑doo‑da... of phase cancellation as you cycle through them. However, the pattern you obtain is fixed and repeats, suggesting that the maximum offset applied to each voice is predetermined, and the Drift control merely determines how much of this is applied. I am always critical of the ‘static offset’ method of imitating the vagaries of analogue synths, and so it is here; oscillators should drift gently, not be initialised at different but constant pitches.
Since the Memorymode is a careful recreation of the original, warts and all, it emulates the Memorymoog’s greatest strength and its greatest failing: distortion. It’s ridiculously easy to obtain three types of distortion from it — what sounds a bit like intermodulation distortion in the mixer (which is not dissimilar to what happens in a Memorymoog), as well as digital distortion in the output stage and aliasing. You can control the first two of these by reducing the oscillator levels and the master volume appropriately, but the aliasing is another matter. With an 8’ sawtooth oscillator this kicks in at around MIDI C6, which is much lower than I had expected.
So I hooked up my signal analyser to have a look at it. What I discovered was odd. Although the aliasing is ever‑present at high pitches, the frequency responses of the oscillators change as you adjust their levels. At low levels, all three oscillators exhibit a dip of around 40dB at 17kHz and higher. At higher levels this dip diminishes to less than 10dB, but with a slight hole at around 17kHz. Yet the frequency response of the noise is smooth at all levels, which means that this is an artefact of the oscillators themselves rather than the filter or anything else in the signal path. Does this matter? The odd oscillator spectrum probably doesn’t, but the aliasing does because, if you program bright Memorymode sounds with high‑pitched components (say, a 2’ oscillator mixed with 8’ or 16’ oscillators, or patches that utilise audio frequency cross‑modulation) you can obtain some unpleasant results. Nevertheless, the Memorymode still performs a creditable job of emulating the Memorymoog if you avoid this and, with a few tweaks, could do even better.
Perhaps I should finish this with two of my favourite aspects of the Memorymode. Firstly, it generates no unwanted buzz or background noise. The noise generated by my Memorymoog has never been quite loud enough to stop me from using it, but the lack of it is lovely. Secondly, there’s no large and noisy fan. While the acoustic noise generated by my Memorymoog can get lost on a large stage, I have always hated using synths with fans in the studio. I’m willing to forgive the Memorymode a lot for these two improvements!
Despite some shortcomings, the Memorymode is a very nice‑sounding soft synth. Does it sound precisely like the instrument on which it’s based? No, it doesn’t, in large part because it’s too well behaved. But it often captures the essence of the Memorymoog, it’s easy to use, it offers expanded polyphony, the extra facilities are well chosen, and it costs about half a percent of what you’re likely to pay for a working original. Even if you bought a dedicated laptop to host it plus a high‑quality audio interface and a 61‑note MIDI keyboard controller, you would be unlikely to spend 10 percent of the cost of a 40‑year‑old lump of silicon, metal and wood that might last another 40 years or (more likely) go ‘phut’ tomorrow.
Although it isn’t obvious from its panel, the Memorymoog offers nine arpeggio modes: up, down, alternating and as played, all either latched or unlatched, plus a chord repeat mode called Auto Trigger. To these, the Memorymode adds a random mode, either latched or unlatched.
Happily, the soft synth’s arpeggiator has its own clock with a range of 0.25 – 30 Hz (the original used the LFO as its clock) but it can be sync’ed to MIDI clock if desired. Furthermore, it offers a range of up to four octaves, which is a welcome addition.
MIDI Learn is nicely implemented; just right‑click (or its equivalent) on a Memorymode control and adjust a connected hardware control to marry the two. OK, that’s pretty much par for the course but, unlike many soft synths, you can direct a given controller to multiple parameters simultaneously. You can also assign keys as controllers, which means that you can do clever things such as switch portamento on and off when you play a particular key, or change its rate, or alter the oscillator mix or... well, the possibilities are huge. You can even do these things using the on‑screen keyboard and the host’s QWERTY keyboard, which could be useful if you’re running a pre‑existing sequence and have no MIDI controller attached. This is also where you assign aftertouch to multiple, simultaneous destinations. (Yippee!)
Having assigned a MIDI control, you can choose its range and whether it affects the parameter linearly or with user‑definable convex (log) or concave (exp) curves. You can even invert its action by making its minimum value greater than its maximum. Since a control can affect multiple parameters simultaneously, with each destination having its own minimum, maximum and response, the possibilities are again enormous. Furthermore, the synth’s relatively small number of voicing parameters means that even a compact MIDI controller can become a powerful programming and performing tool.
Although it lacks the ability to respond to polyphonic aftertouch, the Memorymode is MPE compatible, although only three dimensions are recognised (pitch, channel pressure and timbre) and only a limited selection of destinations is available (see table). Furthermore, the list doesn’t include things such as the LFO or effects because only per‑voice parameters can be controlled in this way.
With MPE enabled, channel 1 is used for standard MIDI messages and channels 2 to 15 carry the controller data, but creating an MPE configuration is no more difficult than creating a MIDI CC map; just right‑click (or its equivalent) on the control of interest and choose your assignment. Having done so, you can edit the selection and the response curve in the same way as before.
|Oscillator 1||Frequency, pulse width|
|Oscillator 2||Frequency, pulse width|
|Oscillator 3||Frequency, pulse width|
|Mixer||Osc 1 level, Osc 2 level, Osc 3 level, Noise level|
|Filter||Cutoff frequency, emphasis, contour amount, keyboard tracking|
|Filter contour||Attack, Decay, Sustain, Release, velocity amount|
|Amplifier contour||Attack, Decay, Sustain, Release, velocity amount|
|Voice modulation||Osc 3 amount, filter contour amount|
- It can sound warm and involving.
- It’s simple to program and use.
- The 12dB/oct filter mode adds a whole new flavour.
- It offers a maximum of 16 voices rather than six.
- It offers flexible MIDI control.
- It’s quiet — lacking both electronic background noise and the annoying fan of the original.
- It’s very affordable.
- It lacks polyphonic aftertouch.
- The on‑board effects are for emergencies only.
- Aliasing is too apparent.
- The online help needs updating and then turning into a manual.
Soft synths have evolved hugely in the past few years, and this one combines immediacy, simplicity and a warm and engaging sound. There’s still work to be done before it can receive an unequivocal ‘thumbs up’ but, at the price, I still view it as a bit of a bargain.