Why do some digital synths sound better than others?

Soft synths sound great. Until you compare them to hardware. Even digital. Any synth by the big three, Yamaha, Roland, Korg, take the synth, find the relevant obscure menu, and figure out how to off all the layers of effects. And what’s left underneath often sounds 2D and like a softsynth.

But then you pick up a classic digital synth. The classic Yamaha FM synths, for example, have huge punch and dimensionality to the sound. When you modulate an operator by another, even on a 2op sound, it feels like you can feel it in your gut. Or the Ensoniqs, like the ESQ-1 and SQ-80. Even if you don’t use the filter, even if the sound lacks all punch and sounds distant and brittle, the phasing between the oscillators when you detune sounds so rich, and like nothing analog could ever do.

Then you do the same thing on a modern synth by the big three. And it all just sounds the same, like software relating to software, not like distinct oscillators crashing against each other. But why?

I recently played a Korg Z1, Kawai K5000s, and Korg Kronos in various music stores. And without the effects, they all sounded 2D to my ears. But the lowly Korg 707 (an FB-01 chip) and the Preen sound HUGE to my ears in comparison. But I just don’t understand why.

I know older Yamaha FMs have 12 bit resolution, and this makes them warm, noisy, and fuzzy. I know clock rate impacts this with lookup tables in relation to aliasing. I know op-amps impact things, and that every time the signal goes into the analog realm this likely impacts things too. I’ve been told bandwith impacts ‘punch’ (ie: the ability to go from low to high volume with clarity very quickly), and maybe more recent synths are sacrificing this and just privileging the processor. I know many more recent synths draw a LOT less overall amperage than classic synths. But I’m no techie. I’m curious to learn, though.

I just know that when I listen to classic Yamaha or Korg FM, or the Preen, they sound huge. And then the Korg Kronos, the Z1, the Yamaha Dx200, the Kawai K5000s, urg. Strip away all the effects, and I hear a softsynth. Why? What’s the technical causes to what my ears are hearing?

Any kind of corner cutting in earlier synths. Aliasing, distortion, quantization noise, hacks and approximations and lack of interpolation in the synthesis operations.

The primary purpose of modern workstations is to recreate well a variety of acoustic timbres. Their synthesis code and the hardware is designed to be as clean as possible.

The upside is that we can still get lovely punch in the Shruthi/Ambika 8-bit oscillators :slight_smile:

@Bjarne More so with the Shruthi oscillator- which is 8 bit PWM, as opposed to the Ambika oscillators, which use 12 bit DACs.


@toneburst Sure, but the oscillator resolution is still limited to 8-bit output in the range of 0-255.

@puggle: I’d be curious to read your opinion on the P6.

I like early samplers because they actually can sound different from each other. When you don’t use any multi samples, the extremes of stretching the sample really shine through. The modern groovesizer TB2 can also take you to the extremes of the sampling spectrum despite running on a stock arduino.

When you think about it the ESQ, Prophet VS, PPG Wave are all samplers with no way to upload custom waveforms.

The three synths you mentioned definitely take some time to get used to. The Korg Z1 is pretty much an older MicroKorg with more features and oscillators, but the basic sounds are very similar. The only things that stick out to me on that one are the three or four modeled oscillators. The K5000s can sound great without the effects, but it is a time sink to learn how to program, but once you get it, you can do anything. Also, if you can, go back to it and sweep the filter with full resonance. You will need to re-trigger a key when you change the resonance amount, but the cutoff is real time. I can guarantee you that only a k5000 filter can sound like a k5000 filter. The Korg Kronos has several revisions, the latest ones have very good orchestral sound libraries built in. if you need a portable symphony, you can’t go wrong.

Since you like the FB-01, I have two as well, I would reccomend you get a TG or SY series synth. Then you can expand your FM sound with samples and vector synthesis. My Tg33 has never failed to impress me, and sometimes the randomizer makes better sounds than I do.

Here is a K5000 demo, and a Tg 33 demo from a small contest we held a while ago.



Tip: The sound you get out of a synth is not the finished result.

So it’s really the imperfections that are making older digital, and the DIY stuff, tend to sound good? I figured it was likely either cheats in the coding, or bits of distortion that creep in whenever you go into the analog realm. It makes sense on the Shruthi and Ambika, but that’s where the Preen confused me, cause it’s got like 18 bit DACs. While I haven’t a/b tested them, I did spend some serious time programming a tg-77 the other day, and its fm sounds much punchier than software to me, it sounds like hardware, albeit more refined than some of the older FM stuff. The Preen is infinitely easier to program, however, with a little more shimmer to the sound, but I haven’t compared them a/b either.

‘Punch’ is I think a function of bandwith, right? The ability to go from extremely loud to quiet quickly with clarity and over a great diff in frequency, yes? This sounds to my ears to be one of the big things the digitals from the big 3 seem to lack, though not their classic offerings.

I tend to like to record my synths pretty dry to let the sound of the synthesis itself show through, so I’m not a big fan of internal effects unless they are as strange and quirky as the synth itself. If I did more ambient music, this would make the k5000 an obvious choice, because it’s effects really are an instrument to itself. Didn’t get to play the Z1 long enough to really say, though.

Never played the Audiothingies P6, I didn’t really like some of the demos I heard, but there aren’t that many out there. But I also tend to not be into virtual analog. I like my analog to be really analog, my digital to be really digital (hence FM), and my hybrids to be strange and hybridy, but I’m not a fan of something trying to be what it’s not (ie: reproducing acoustic sounds, or digital trying to be analog, etc.).

Please be advised that what you hear is always as much, if not more, the result of your expectation than it is the result of the sound that reaches your eardrums. This is why any kind of serious testing involving audio is always done double-blind and why audiophiles almost always refuse to participate in double-blind tests. :wink:

> ‘Punch’ is I think a function of bandwith, right?

No. Bandwidth is the range of frequencies the converter can output, it has little to do with “punch”.

What you call punch is probably due to the shape of envelopes, but I suspect what’s at play is also loudness (the most important factor in the perception of audio quality). Synths with fewer polyphony voices output the signal at a higher amplitude than synths with large polyphony (which have a fairly low level and give you more “headroom” to play new notes on top of the currently playing note). This simple fact make them appear “stronger” and more “solid”.

This is somehow slightly related.

Could the harmonic distortion caused by older/lower bit rate dacs also be a factor?