[mrdos]

Originally Posted by MBTC

I think what Timo's example illustrates though is that a digital synth can emulate just about anything an analog synth can do (with regard to what the human ear would be able to distinguish). Even to the point of getting somewhat randomized "flavor" to each oscillator to emulate what happens with unstable voltage controlled behavior.

Whether they do or not in practice is really a different story, because of course those kinds of features actually end up requiring a lot of processing power - this is one of the reasons newer analog modeling synths gobble up a lot of CPU.

Believe it or not, the Freescale DSPs that are in the Virus are not particularly powerful. In fact they are so yesteryear that I believe that lack of innovation in the chip line might be one factor holding back the Virus product line from moving forward. Supposedly Christoph Kemper cut his programming teeth on these chips (formerly Motorola 68 series), so basically he is deeply invested in them. This is perhaps one reason he decided to work on new audio related products that are based around this chip rather than to explore using alternative chips for the next generation of Viruses.

When I say they are not particularly powerful, I only mean that the CPU that is in your PC or Mac (hell, probably the chip that is in your phone) is capable of much more (including floating point operations). Now a synth DSP does have a lot of advantages, for one thing it does not have to juggle music making while running a general purpose operating system that might be also checking your email or refreshing web pages at the same time, and that greatly simplifies the tasks it has at hand. Also, I remember reading that these Freescale chips have specialized parallel filter processors, which might help explain what many regard as the Virus' most noticeable characteristic (fast/aggressive filters), and some could say that these chips are "designed for" audio algorithms in the sense that they may do things more efficiently from a power consumption standpoint, but in terms of overall processing power they are no where close to something like an Intel Core i7 or a modern gaming GPU.

If there is something holding soft synths back, it may be an over-reliance on mediocre open-source algorithms. I think sometimes plug-in makers may not have the background or knowledge to write all the audio processing code from scratch (nor time to learn it all), so they fall back on the same previously-written open-source algorithms that every other "quick and cheap" plugin maker is using, which often results in a dull, "same old" sound syndrome. This is why some plug-ins just seem to bring a better sound to the table, more talented programmers, more specialized algorithms, and just more time spent toward a finished product. It took the Virus many generations and years to get to where it is today, and I don't know how many plug-in makers are going to put in that level of dedication when a plugin only sells for a couple of hundred bucks a copy, one copy gets sold for every 500 copies that get pirated, etc.

Honestly there are times when I'm running the Virus side by side with Dune2 and I wish the Virus could sound as rich as Dune2. That said, Dune2 can gobble up a huge amount of CPU with a rich unison patch, and there are certainly some things the Virus does better.

I agree with you for sure on a lot of these points, but I would add a couple of things for food for thought.

Sounds like you know your stuff on the tech side; and I have heard that story about the chips and the whole thing about Kemper wanting to stick with Motorola, etc. That might be true. But, I have some other thoughts.

I'm a software engineer (have been for about 20 years); I have seen some papers that suggest that specialized audio DSP chips are still superior to X86 when it comes to processing audio, and that is not on a mhz to mhz level. But I could not find many, when I looked some time ago. I don't know the speed or power of these chips in particular, but you can sort of do a comparison here... Take a very fast X86 PC, run instances of Serum on it, turn the unison up (to make 8 voice hyper saws), and start adding more instances and voices (or do the same thing with Massive). Most PCs will quickly crap out before the limit of the Virus, even Core i7 ones, and that is, as we know, with for the most part inferior fidelity. Check out https://www.youtube.com/watch?v=JR-3Bn3Poz4 re voices. I have heard some great components here and there (for example I think Serum has great filters, but the effects suck, and the oscillators can be very useful, but sometimes lacking to my ear, in the mix; Massive on the other hand has very solid oscillators, pro-mix wise, but the filters suck–the Virus has it all, all the time).

Think too that when the Nord Modular G2 came out, arguably one of the most complex digital synthesizer design projects of all time, there was a software version demo, which could produce basically the same sounds, but limited to a single voice. I don't know if that says anything, but to me it says that these companies may be able to translate their stuff to X86, perhaps in draft form (unoptimized), but choose not to, because it would break their business model (and probably still because there is something to the hardware DSP sound that simply cannot be emulated, even if it is related to the components/converters).

But I agree that whether synths do this stuff in practice is all subjective and varies widely. I think, personally, one of the big selling points of the Virus is, as you said, software hasn't caught up in some areas... And I agree, I think it is for the same reasons you think it is.

I think there is an over reliance on open source frameworks. This contributes to all the plugins sounding, as the producer Flood has said of software-based audio, "samey". And that is my biggest criticism of software stuff. It all sounds kind of the same at its core; even when it's the most different. Hardware seems to be totally proprietary, and that is why hardware stuff are real instruments that can last forever.

(https://www.youtube.com/watch?v=FZRBJNTCl9c skip to the very end; interesting A/B comparison of the same patch made in Sylenth and the Virus... No comparison).

I too think piracy really hurts plugins. Some years ago I stopped using anything I didn't buy, and that has improved my production; I always am at my best when I limit myself to a couple of plugins and learn them deeply, rather than downloading something new every week. So for me that's Serum, Massive, my Nord Modular G2, and the Virus. Oh, and I bought U-he Bazille, too. Those make me super happy Though the Nord frustrates me

I argue that DSP is superior for audio for various reasons, but mainly it is evident in the sound. It is more than just software running on a different platform, though software running on a *better* platform is part of it.

If you consider, for a moment, a couple of things:

1) Power efficiency: look at your iPhone/Android phone. That processor is a different architecture, and the clock speed is less than half of what your desktop/laptop is. It is optimized for a different use of power. It also heavily utilizes DSP chip(s) to handle the audio transactions... Why? Why not do it all in the main CPU? Because the DSP chip is so much more efficient per clock cycle, it saves a lot of power to do it that way. But that DSP chip is nowhere near the clock speed of the CPU.

2) Closed platform optimization: Again taking the phone as an example, the other part of the reason the iPhone in particular was so snappy and instant-fast early on (the first couple of iPhones had quite weak processors and little RAM) was that developers were heavily forced to optimize their code. I develop iOS apps, and this is actually still the case, though a lot of things have been eased as the CPU and RAM have grown greatly in later generations.

Another example is console gaming machines. Like Playstation/XBox. If you look at the lifecycles of these machines, relative to their PC counterparts, the same cross-platform games that come out on all platforms will require greater and greater specs for the PC version (because developers do not optimize the code; instead they let consumers buy faster hardware–it would not make financial sense to allocate resources into further optimization when consumers have demonstrated they will upgrade). Yet the console versions will, for the most part, look the same, due to increasing code/hardware optimization. and over the span of the console's life, as open and closed-sources libraries and optimization techniques improve, the graphical fidelity/performance will basically reach the level of the next generation of consoles when those are released. Hopefully that makes sense.

So my point is just that even though some of these synths may have old hardware, it may not be apples to apples. There is some objective stuff like voices/hypersaws that can be compared, which I think has merit, but there is a lot in the subjective sound quality area, as well, that can't really be compared... Though I would say that really, among other things, the filters, for example, of the Virus, are still unmatched by any other digital synth. Is that worth it to some people? It will be to some, and not to others. Can the listener tell? Some can, some can't. What is music, anyway?

Anyway, long post, but I think the idea that hardware should go away... Well, I think hardware still has a specific sound and purpose, and while I only have ideas as to why, it's all certainly very interesting to me.

I will say, I think they don't really need to change anything about the Virus, really... Sure, additional synthesis options would be great, but I think they would sell more if they would just improve on the TI concept... Eg, maybe Thunderbolt, or USB3, and transmit 16 simultaneous outputs, so you can just use the whole thing without having to record individual tracks, etc. Do that and add one new oscillator option or whatever, and you have the TI3! Done!