Software Synths vs Hardware Synths? They each have their pluses and minuses and here is my experience with each and why I've decided to go with a mix of both in my studio:
With all of the advantages that software synths may have, there is ONE HUGE disadvantage: They are resource hogs! Hardware synths don't face any resource issues. They are self-contained units that don't require any particular memory or CPU specification to help them along. Theoretically a hardware synth, as long as it doesn't brake down, can run indefinitely without requiring an upgrade to any of its parts. Not so with software synths. These need as much processing horsepower as possible to make them run efficiently.
On the other hand, a major advantage of software synths is that as long as you can keep increasing processing power and memory you should be able to surpass the polyphony(by stacking patches on different channels) and effects processing power of hardware synths. For instance, say we were able to buy machines, without breaking our wallet, that offered 12Ghz CPUs and 12 Giga bytes of memory. There is no telling how many instances of any given plug-in can be instantiated. Hardware synths are a closed boxed. Once the memory and polyphony are maxed out, per manufacturer's limits, that's it! The polyphony and effects processing of hardware synths are fixed forever.
All that being said, a phylosophical question beckons, namely: how much is enough? In other words, what QUALITATIVE difference can the human ear discern between, say, a Korg Trinity's resonant sweep patch and a similar one created by Absynth? Doesn’t it boil down to a matter of personal taste? Granted, there are some bad synth engines out there that produce horrible timbers, but when we start comparing the very best hardware synths to the very best software counterparts, there are very few important audible differences. The differences are easier to discern when we compare a piano sample libray that's 40G bytes large vs one that's a few hundred megabyte. For sampling applications that specialize in acoustic instuments, software wins hands down. I can’t imagine anyone arguing otherwise with any success. But can the same thing be said for software plug-ins that specialize in synthetic sounds?
Someone will argue that software synths, given enough processing power and memory, can produce larger than life sounds because of their ability to stack larger samples, each one with its own distinct characteristics. A case in point would be a comparison of the original Korg Wavestation SR with its software clone. The latter, arguably, is more powerful since it can be instantiated as many times as resources allow with the goal of creating LARGER, DEEPER sounds. But, the same question is asked: when is enough, enough? Where does sound become unattractive, or even unpleasant? We can, theoretically, stack an infinite number of basic sounds. Is that where synthesis is heading? Following the muse that more is better?
We currently have tools at our disposal like Absynth and Spectrasonics Atmosphere. These plug-ins can morph sound into shapes that will raise the hair on the back of our necks. However, when anyone reviews, or better yet, uses these synths on a real world project, you will hear the same story: "These are great sounding tools, but are limited to certain specialized applications: soundtracks, sound effects and sound design." Is this what the vast majority of the public is doing? There’s probably a larger group of artists experimenting with electronic sounds today than at any other time in the history of synthesis. But does that guarantee that we’ll be hearing better music? Some will argue that is the case. Yet, some will say that we are only hearing convoluted noises instead of musical notes.
All the pilosophical issues aside, there is one area that seems to be consistently overlooked when considering soft vs hard synths: the all too important "MIDI program and bank changes". Seems that very few, if any, people using software synths run into this limitation, or care to mention it when discussing this topic.
Take my case: my pieces tend to be longer than 5 minutes; usually 7-9 minutes long. For songs that long I use LOTS of different patches and combinations(20-40 distinct patches is about average for one of my pieces). When using soft synths I cannot change programs or banks via MIDI(not on Dimensio Pro or Arturia CS-80V). This limits the number of soft synth instances I can run for a given piece. Anything over six to eight instances overloads my CPU.
For example, when I run Ivory Grand I can only run one instance of it(not that you need to run more than that) and MAYBE another two or three instances of any other plug-in synth. And I'm running a 3GigHz, 2 Gig Byte machine! In order to run more soft synth instances I have to run them on my other PC. But that is not the answer. My second PC is the one that runs Sonar 5 and where I record my audio. So running too many plug-ins there will tax the resources of that PC.
On the other hand, with hardware synths I don't have this problem. I simply change program as often as I need to. I don't have to worry about CPU overload or memory consumption. Having four hardware synths I have plenty of combinations or patches running simultaenously and swap them out via MIDI program change parameters. Some folks don’t like to work in this way citing that it’s a headache to keep track of all the program changes. In my case I see it as a way to not have to upgrade my PCs with each software synth I add. Its’ really a matter of enconomics for me.
Here is wha I currently run in my studio:
Korg Triton Rack
Kork Trinity Rack x 2
Korg Z1
M-Audio Keystation Pro 88
Arturia CS-80V
Synthogy Ivory
Cakewalk Dimension Pro
Cakewalk Sonar 5
Dell Dimension 2.8 GHZ and 1Gig bytes RAM
Dell Dimension 3 GHZ and 2 Gig bytes RAm
I find that with this configuration I can use my three plug-ins as they were intended, without any hicups.
I design and compose my music in such a way that I know the limits of my software and don’t come close to reaching its peformance boundaries. My hardware synths are my workhorses and until computers with more power and memory become affordable I will hold on to them.
A development that may convince me to drop my hardware altogether would be if sofTware synth vendors were to incorporate MIDI program/bank change capability into their products. I see this as a very difficult thing to implement since most of these tools use large samples that reside on disk and to swap them in and out of memory would require immense processing and bus I/O speeds. An obvious solution would be to load all of the samples into memory once the plug-in is instantiated for the first time. This would make it easier for the host sequencer to swap out samples as needed. However, this still points us to larger memory(in the teens of Giga bytes) and blazing fast CPUs. Will these be available soon and, more importantly, at an affordable price?
For now, hardware synths are a viable solution to the restrictions imposed on computer resources by software synths. Hopefully, in the near future we’ll all have that 12 Ghz, 12Gbyte monster sitting in our studios.
JR Fernandez
January 9, 2006