Cherry Audio Releases Mercury-4 Compuphonic Synthesizer!

[Friendly Noise]

Level matched version of the first video. The old one will be deleted:

https://youtu.be/REm2HpePj7o

[briefcasemanx]

Level matched version of the first video. The old one will be deleted:

https://youtu.be/REm2HpePj7o

Maybe I'm insane or hearing things, but it sounds like there's something interesting happening on the initial attack of the analog synths single notes. Like a timbre change or something that is more pronounced on the analog.

Looking at the audio in my DAW the analog is doing some very weird stuff. In the beginning when you're doing saw waves with no chorus or saturation, the analog waveforms have a strong DC offset (not sure if this is technically the correct engineering term, but the basically the positive peaks are way lower in magnitude then the negative waveform peaks, making the entire waveform look as if it had been shifted downward). Weirdly, the the positive peaks, which have a small magnitude, it almost looks like there's some clipping. The peaks are mostly shaved off at certain value, it looks pretty flat. The negative peaks, even though they are lower in value and thus should be closer to clipping, have random values -- they don't look like they've been lopped off at the same value.

jupiter 4 vs mercury 4 waveforms.jpg

It looks as if earlier in the circuitry there was a DC offset in the OPPOSITE direction than is shown here, where positive values were much higher in magnitude, then the signal got clipped, then later in the circuitry something caused a DC offset in the other direction. The amount of positive half "clipping" visible seems to change with note pitch too. There also seems to be microfluctuations in timbre that I don't think is just purely noise.

Someone a few weeks ago said that emulating an analog oscillator is as simple as using a simple wavetable. I don't think this is true at all, unless maybe it's some insane multi-dimensional array of wavetables with some sort of RNG implementation. I don't think anyone who thinks the oscillator portion of emulation is simple is ever going to make a convincing analog emulation, but then again, maybe no one will.

[XpanderDude]

Anyone who thinks it's simple is an example of the dunning-kruger effect in action. Especially as you go back to these late 70s synths with pretty wide tolerances, you start to see stuff (like you posted above.) Actually, that's a pretty good observation you made - and it's like - how do you model this?

Granted, sometimes you don't want clipped and wild tolerances per voice, but that's one of the hallmarks of the Jupiter 4 and what some people really do want.

[Friendly Noise]

Level matched version of the first video. The old one will be deleted:

https://youtu.be/REm2HpePj7o

Maybe I'm insane or hearing things, but it sounds like there's something interesting happening on the initial attack of the analog synths single notes. Like a timbre change or something that is more pronounced on the analog.

Looking at the audio in my DAW the analog is doing some very weird stuff. In the beginning when you're doing saw waves with no chorus or saturation, the analog waveforms have a strong DC offset (not sure if this is technically the correct engineering term, but the basically the positive peaks are way lower in magnitude then the negative waveform peaks, making the entire waveform look as if it had been shifted downward). Weirdly, the the positive peaks, which have a small magnitude, it almost looks like there's some clipping. The peaks are mostly shaved off at certain value, it looks pretty flat. The negative peaks, even though they are lower in value and thus should be closer to clipping, have random values -- they don't look like they've been lopped off at the same value.



jupiter 4 vs mercury 4 waveforms.jpg



It looks as if earlier in the circuitry there was a DC offset in the OPPOSITE direction than is shown here, where positive values were much higher in magnitude, then the signal got clipped, then later in the circuitry something caused a DC offset in the other direction. The amount of positive half "clipping" visible seems to change with note pitch too. There also seems to be microfluctuations in timbre that I don't think is just purely noise.

Someone a few weeks ago said that emulating an analog oscillator is as simple as using a simple wavetable. I don't think this is true at all, unless maybe it's some insane multi-dimensional array of wavetables with some sort of RNG implementation. I don't think anyone who thinks the oscillator portion of emulation is simple is ever going to make a convincing analog emulation, but then again, maybe no one will.

While your comments about the waveform are very interesting, please don’t forget that the YouTube conversion might’ve changed the original sound somehow. A few weeks ago I uploaded to my channel a sine wave test to test low end. The sound in the YouTube video was completely different and rendered the test unusable. I had to link the unprocessed sine waves or the test would be a waste of time. On the other hand, the test proved how dramatic the YouTube conversion can be.
https://youtu.be/N_Zy4-26htM

[Friendly Noise]

Part Three:

https://youtu.be/l09ZxnofSmw

[wwjd]

thanks for rematching the levels on that first video, well done. I'm even more impressed with the software now.

as an owner of a real Jupiter 4, I'd love to hear your unfiltered, subjective, competely opinionated perspective of this software version. I liked it and already bought it, not like I am gonna return it or anything, just would love to hear an owner's perspective.

[Friendly Noise]

thanks for rematching the levels on that first video, well done. I'm even more impressed with the software now.

as an owner of a real Jupiter 4, I'd love to hear your unfiltered, subjective, competely opinionated perspective of this software version. I liked it and already bought it, not like I am gonna return it or anything, just would love to hear an owner's perspective.

There is still a fourth and last video coming in the next days with monophonic sequences. After that, I will write down my opinion about the Mercury 4.

[Friendly Noise]

The fourth and last part:

https://youtu.be/R49V5GvIPA8

[Friendly Noise]

My opinion about the Mercury 4 so far:

Did Cherry Audio did a good job?
Yes, no doubt to me. And they are the first company to bring a Jupiter 4 emulation to the market. That deserves a great respect. Why? Because the Jupiter 4 is to me one of the most idiosyncratic synths ever made. It’s full of character and quirkiness, despite of the fact that the synth engine seems to be nothing special at first sight. That’s the reason because the Jupiter 4 is having a renaissance in the last years and it’s becoming one of the most sought after vintage synths. Second hand prices reflect that.

Why is the Jupiter 4 so special?
You have to listen to it to answer this question? In fact you have already listened to a Jupiter 4 in lots of very well know and famous songs from the 80’s. While the sounds featured in these songs showed mostly the “standard” use of the Jupiter 4, there is also a more experimental side of the synth, which will be obvious when you crank up the VCA Slider and the sound saturates like hell, or when you use the amazing Roland filter and modulate it and/or the pitch with the LFO in audio range. The sound of the square wave (and also of the saw wave to a certain extent) are quite special, too. As I like to say: not all square waves were created the same.

Did Cherry Audio model all the features?
Yes, sometimes very good, like the basic sound of the waveforms. The VCA saturation is there, but it doesn’t kick in so smooth and, at maximal level, it’s not so strong like on the hardware. In fact, the saturation in the Jupiter 4 VCA stage can be so strong that it could sometimes render the sound unpractical for standard purposes.
The filter sounds good, but with high levels of resonance it lacks some smoothness (you can clearly hear the bumps created by high levels of resonance in the long release examples of Part Three).
The curves of the envelopes is a masterwork on most Roland synthesisers. They are musical and usable, not matter the settings. On the Mercury 4, the curves are noticeably different. I don’t mean the times, which are also different, but the curve value for each envelope segment. So, if you don’t have the reference of the hardware, envelopes are ok, but I had sometimes to struggle a little in order to match the sounds.
The very fast LFO is also in the Mercury 4. I think the LFO goes even faster than on the hardware, what can be only welcome.
The key modes are faithfully reproduced, as far as I could test. With the added convenience of the software: more polyphony or monophonic mode if needed…
I didn’t test some functions like the arpeggiator. So I can’t comment on that.

Did I missed something on the Mercury 4?
Yes, and it is a key feature that may explain some of the differences in the examples: there are trimmers on the four voice cards of the Jupiter 4. By tweaking the trimmers you can adjust crucial parameters like tuning, filter cutoff bias, resonance amount, attack time, etc… per voice. That means: chords in a Jupiter 4 can sound very tight or from slightly to heavily detuned, depending on how you tune the voice cards. The tuning of the four voices also decides how the unison mode will sound.
In my case, I tweaked the attack time for a longer value. And slow pads now sound marvellous on my Jupiter 4. When I get it… not so great. I noticed different attack time on the Mercury 4, and to my ears not so nice as with the hardware. But it is because I tweaked the voice cards, because Cherry Audio didn’t choose the right values for the envelope, of even because of both reasons???!!!
Anyway: individual control of sound parameters per voice is a key feature in the real behaviour of the Jupiter 4. Without that, you can only come so close to the sound of a exiting Jupiter 4.

After all that, should you get the Mercury 4?
To me the answer is yes. If you like the Jupiter 4 and you don’t have already something that can sound close, then the Mercury 4 can get you close to that sound, and it also gives you the usual convenience of the software. I haven’t say anything about the included delay and reverb, because the hardware had not effects at all, but the standard effects are no doubt a useful feature.
In a real recording, there is little chance that you would listen exclusively to the raw sound of a instrument. And there are enough great plugins today to bring the sound of your raw recording closer to what you have in your ears. For that reason, I didn’t use any processing to enhance the sounds or trying to get both instruments to sound closer to each other.

What did my test videos proof?
Nothing more, but nothing less that a real comparison between hardware and software. It’s not a scientific test, so results may vary in further tests made by other people or even by myself.
There are some elements which may alter the output of the test:
a) My Jupiter 4 is not the holy grail of the Jupiter 4 sound. Other hardware units may sound better, worse, or just different to everyone’s ears.
b) My Jupiter 4 was never callibrated/set up/recapped since 20 years. Does that affect the sound?
c) Mercury 4 is not modelled after “my” hardware unit.
d) Mercury 4 has not control over individual voices so far.
e) User error is always a possibility. This kind of test is long and tedious. I can get tired and make mistakes after some time or maybe not invest enough time to match the sounds. Other tester could succeed where I didn’t.
It’s only fair to make clear that sonically different results can be caused by a combination of several circumstances.

Obviously, the great price asked for the Mercury 4 is a reason to get it. I don’t think that a fair price is an excuse for unsatisfactory quality. But in this case, I do think that Mercury 4 is good enough, considering that it is the first available Jupiter 4 in software, and specially considering the real challenge of perfectly emulating a Jupiter 4.

[rezoneight]

Nice conclusions and absolutely fair. Thanks for the comparison!

[Mr Arkadin]

I didn’t test some functions like the arpeggiator. So I can’t comment on that.

One of the key features in my opinion. The manual states that it follows the quirky arpegiattor function and it does seem to capture that difference in sound I always noticed with other arps.

For me, apart from maybe an under-the-bonnet function to tweak the cards like some synths provide, the only things I'd like to see are:

1. The "non-arp arp" trick of holding down arp+poly (or mono) and hold buttons (I think) where you can have a pulsing note or chord without it arping up and down.

2. A virtual trigger input (would be MIDI from DAW) where you could trigger the arp via a different MIDI channel (like a side-chain) so that you can say use a snare to trigger when the arp start. This allows for odd time signatures, funky arps etc.

[wwjd]

great review from an owner!

[Friendly Noise]

I didn’t test some functions like the arpeggiator. So I can’t comment on that.

One of the key features in my opinion. The manual states that it follows the quirky arpegiattor function and it does seem to capture that difference in sound I always noticed with other arps.

The arpeggiator in the Jupiter 4 is not the usual arpeggiator found in other instruments, but I wanted to test only timbral differences. While the arpeggiator may be a funny guy, it doesn’t affect the sound engine, afaik.

[Mr Arkadin]

The arpeggiator in the Jupiter 4 is not the usual arpeggiator found in other instruments, but I wanted to test only timbral differences. While the arpeggiator may be a funny guy, it doesn’t affect the sound engine, afaik.

Sure, it doesn't affect the timbral character, but it does affect the overall character of the synth. I was thinking in terms of people saying the UVI sounding better because it was samples, but I bet they have given it a standard arpeggiator (especially as it covers the other Jupiters too). Nice tests though so no criticism here.

[Friendly Noise]

The arpeggiator in the Jupiter 4 is not the usual arpeggiator found in other instruments, but I wanted to test only timbral differences. While the arpeggiator may be a funny guy, it doesn’t affect the sound engine, afaik.

Sure, it doesn't affect the timbral character, but it does affect the overall character of the synth. I was thinking in terms of people saying the UVI sounding better because it was samples, but I bet they have given it a standard arpeggiator (especially as it covers the other Jupiters too). Nice tests though so no criticism here.

I’m sorry I didn’t try the arpeggiator. It really affects the overall character of the synth, exactly as you said. I didn’t want to make the test any longer. For example, the included reverb and delay in the Mercury 4 are a nice bonus, but those can be tested just by downloading the demo, without A/B comparisons with the original hardware.