Quantcast
Why could some digital hardware synths sound better than software? - Page 2 - Gearspace.com
The No.1 Website for Pro Audio
Why could some digital hardware synths sound better than software?
Old 10th September 2012 | Show parent
  #31
Lives for gear
 
🎧 10 years
Quote:
Originally Posted by JimmiG ➡️
I think the biggest reason is R&D.
That used to be quite of an issue, not any more. Sean Costello of Valhalla DSP has forgotten more 'bout reverbs than most (note: most, instead of all) people in hardware shops will ever learn.


Quote:
Originally Posted by Michael E ➡️
By the way, how do you think - why we don't have any DSP based virtual synths in small package?
In another words like Virus but without controls & converters. Just USB.
Why? TI runs on 2x 56K DSP clocked at 150MHz. New i7 runs about 50 circles around that. And, you would need custom USB driver and latency would allways be higher that pure SW synth. Access have their reasons why they wont port Virus to plugin but it's not related to silicon.

Quote:
Originally Posted by Michael E ➡️
Like hardware video converters which do the work about 5 times faster than just software.
Those things have ASICs inside designed to do exactly and only what they do. You can have HW CODEC that way that runs on 20 MHz that would in software form choke general purpose processor clocked at 700MHz.
Old 10th September 2012
  #32
Lives for gear
 
fastlanestoner's Avatar
 
🎧 10 years
Because electricity moving through a circuit sounds better than 0s and 1s
Old 10th September 2012 | Show parent
  #33
Lives for gear
 
🎧 5 years
Quote:
Originally Posted by fastlanestoner ➡️
Because electricity moving through a circuit sounds better than 0s and 1s
Well besides the fact, that "0s" and "1s" are just our symbolic representations of actual electricity moving through the circuit of a CPU - this thread was about Digital Hardware Synths, which means: Specialized Computers with a specialized CPU called "DSP" running some software.
Old 10th September 2012 | Show parent
  #34
165099
Guest
Quote:
Originally Posted by fastlanestoner ➡️
Because electricity moving through a circuit sounds better than 0s and 1s
which is a valid statement of opinion if this topic were about the difference between the sound of analog and digital synths.

but as the name implies, digital synths (whether based in hardware or software) happen to be digital and digital is composed of zeros and ones up until it reach a Digital Analog Converter (DAC).

it is just one after another on this site...
Old 10th September 2012 | Show parent
  #35
Lives for gear
 
Michael E's Avatar
 
🎧 10 years
Quote:
Originally Posted by Persemone ➡️
We do, don't we? Powercore, UAD? Virus or V-station?
- Powercore project seems just dying.
- UAD has no synths.

Quote:
Originally Posted by recnsci ➡️
New i7 runs about 50 circles around that.
It's on paper. Massive sounds the same not matter what CPU you have.

Let me bring another metaphor.
If you are going to play the best new video games - you have to get a very good video card.
Without it everything will be emulated by CPU and result will be very pity.

Same with the sound.
Yes native stuff may be good, yes it does what it should, but... when you hear VA synth or hardware reverb - it sounds much better.
Old 10th September 2012 | Show parent
  #36
Gear Guru
 
zerocrossing's Avatar
 
🎧 10 years
Quote:
Originally Posted by msl ➡️
Different convertors, the sound of the specific dac.



.
I've heard this brought up a lot, and tbh, it doesn't really make much sense in context, unless you're talking about older DACs which are actually lower fidelity than most decent DACs out today, but desirable for their "crunchiness" probably due to how they use dithering to make the low bit rate sound better.

So you can have an amazing DAC on a digital synth... but doesn't it eventually end up going into your audio interface anyway? This chain can only be as good or bad as it's weakest link. I guess if you're doing it all in tape... but that's a pretty rare thing these days.

I think the actual truth, that's already been mentioned, is that a good digital synth uses 100% of it's power for a singular purpose with no compromise.

This doesn't mean that all digital synths sound better than software though. I've tried out SH-201, SH-32, AN200, MS2000 and Gaias and Microns that don't do as well as modern software. You have to pay up if you want the good stuff, though the old DX's are still a value compared to FM8. I love my DX200.
Old 10th September 2012
  #37
Lives for gear
 
jarlywarly's Avatar
 
🎧 5 years
Quote:
Originally Posted by StarfishMusic ➡️
What sparked me to write this thread Is I recently spent a ton of time playing Native Instruments FM8. Though the FM8 does some crazy stuff the Fs1r doesn't, just the clarity, sparkle and integrity of the Fs1r seems far superior. Both are FM synths, shouldn't the FM8 at least be in the same leauge? It has that same dirty sound character alot of software synths seem to have (not samplers/romplers) I realize the fs1r is a 16 operator synth and the FM8 6 op but still I'm not talking about the complexity of the sounds. Just simple sounds too on the Fs1r sound superior. This makes no sense when the Fs1r converts out of digital from itself and gets reconverted into my software making it second generation. The FM8 stays in the computer and only comes out of my DA. My converters and clock for my DAW have to be better than what can be in a 14 year old synth! Im using an SSL Alphalink and antelope Ocx. They do an excellent job of recording that tight clean FS1r sparkle so it has to be from the softsynths. Anyone have similar experiences?
I also own a FS1r and I somewhat relate to what you are saying. There is definitely something about the DACs on that machine that gives a fuzzy warmth to the output (subjective). Having said that, the FS1r has an additional advantage in it's incredible FX section which, for me, is a defining part of it's overall sound. On top of this, the FS1r has some of the most impressively detailed patch designs I've ever seen. Those EPs are to die for! Creating them from scratch would be totally beyond me and, I suspect, the majority of owners.

I never owned FM8 but I do have several other FM softies: Rhino, Blue and Octopus. They all sound different - different waveforms, filters, etc. Of these, Octopus is my favourite - it's FM engine, to my ears, comes the closest to the raw FS1r FM sound.

Nevertheless I'm glad they sound different - not 'better' or 'worse', just different. Advantage of the softies over the FS1r is the vastly increased modulation possibilities and the convenience of full DAW integration. Swings and roundabouts.

I don't stress about the differences - I'm just grateful to have the choice.
Old 10th September 2012
  #38
Lives for gear
 
🎧 5 years
Unlike analog synthesis methods, we are today technologically able to recreate any hardware digital synth, at least to the point of the DA converter. But potential is rarely actualized, due to constraints on time, knowledge, skill and money.

There may technical aspects of the synthesis implementation that play a factor. For example, classic yamaha FM is chip based, but not designed around a general purpose CPU. The way FM is implemented in the Yamaha uses some shortcuts and approximations due to the components used, that would not be obvious to make if one were implementing FM on a very powerful general purpose computer.

The other, probably more important factor, is that the software is written by people who didn't make the hardware, and that the software is it's own program and not the same as the hardware. Even when the same company makes a digital version for example, the korg legacy digital collection, they still make changes or add features to differentiate the software.

A parallel example would be arcade games. I myself quite like to play Defender. The atari 2600 port is decently fun, it does bear some resemblance to defender, but no one would say it is the same as the arcade. The atari 800XL version is really quite good. But the input controls are different, and the graphics and gameplay, though much more better than the 2600 version, is still below the original. Then you can get MAME which emulates arcade boards down to component parts. Defender on MAME IS defender in the arcade, at least as far as the 1s and 0s go. But when you get to the DA stage, well, the emulated sound chip doesn't really sound the same, nor is playing on a LCD screen the same as on a big CRT.
Old 10th September 2012 | Show parent
  #39
Gear Nut
 
🎧 5 years
I love many soft synths and am not a purist, by far, but I cannot help but recognize that, for instance, my Nord Lead 2X, which is a VA, all conditions being the same, sounds better than any soft synth that I have tried so far, including DIVA.

It is not only the sound of it, but the envelopes responses, filter etc.

Software is great and it does have a place in modern music production, but the day that, "sound" wise, hardware could be entirely replaced by software, IMHO, is yet to come.
Old 10th September 2012 | Show parent
  #40
Lives for gear
 
🎧 15 years
Quote:
Originally Posted by subsonix ➡️
Yeah, I'd agree with most of this.

The same problem seems to appear in all the plugin synths I've tried.

Sure they have the general feel or frequency response to the hardware synth they are trying to emulate (or are influenced by), but there seems to be some kind of resolution factor missing from the plugins which make them soft sounding and cloudy in the mid-range.

It's almost like the coders are emulating the frequency response only and forgetting the dynamics.

It's in the dynamic of the attack that a lot of the magic lies with hardware synths and can make the difference between a tight punchy bass and a round mushy non-entity.

Not on an ADSR scale but something of a much higher resolution.

So I guess even on the digital hardware there is something happening in the output stage that adds some extra magic that coders are either missing, or that chews up so much cpu power that they don't bother with it.

It's not the output stage. This problem persists with native plugins almost since they first came up. Not only synths, it's the same for processing.

There are some plugin synths ported from Linux that use integer math, they sound very clear and nice to my ears. The same for the processing in Sawstudio, which is 100% integer programmed in assembler.

I have a suspicion that there are certain factors in the common plugin formats and the way coding and calculating is done that causes the sound to get dull and liveless.

But lacking the technical insight I long ago decided to not waste time with it anymore, I'm now 100% hardware for all important sound generation and processing (except for mastering), the results are worth the trouble.
Old 10th September 2012 | Show parent
  #41
Lives for gear
 
🎧 10 years
Quote:
Originally Posted by Michael E ➡️
It's on paper.
????????
And I thought native ports of Powercore stuff killed that "pixie dust in DSPs" story long ago...

Quote:
Originally Posted by Michael E ➡️
Same with the sound.
Nope. As in "Only difference with Virus TI algos running on Intel with integer arithmetics would be slight influence of analog path". You could even do 24 bit mask to get spot on and it would still eat less than 10% of i7 at max polyphony. There is reason why Access will not make plugin port of Virus and it has nothing to do with what silicon will it run on.
Old 10th September 2012 | Show parent
  #42
Lives for gear
 
2 Reviews written
🎧 10 years
i still think hardware sounds better then software...

but software are coming closer and closer.. but not yet close enough
Old 10th September 2012 | Show parent
  #43
Gear Maniac
 
🎧 10 years
Quote:
Originally Posted by nznexus ➡️
i still think hardware sounds better then software...

but software are coming closer and closer.. but not yet close enough
same here. I will continue to use my hardware until plugins get to the point where I don't notice that they are flat and lifeless
Old 10th September 2012 | Show parent
  #44
Gear Nut
 
HHaynes's Avatar
 
🎧 10 years
What a hybrid analog synth taught me about placebo effect

Quote:
Originally Posted by nznexus ➡️
i still think hardware sounds better then software...

but software are coming closer and closer.. but not yet close enough
Funny - I'm going to write a long-ish post - but yours pretty much sums things up nicely.

:D

As some of you know, my first 'gig' out of college was working for Bob Moog. Yes, *the* Bob Moog. Aside from the general summer intern type sweep-up tasks, I was handed a soldering iron and started checking out Memorymoog instruments that were to be mod'ed using a design by Rudi Linhard - a genius German engineer that had a real passion for sound.

The Memorymoog had a lot going for it - and several strikes against it. On the sound level, it used the SEM audio chips for VCO and VCA (maybe other signal path stuff??) - common for synths at the time and a non-controversial design short-cut by Norlin, the company that bought Moog Music from Moog. It still used a discrete component filter ladder for the VCF, which tended to preserve "the Moog sound" - though there's some debate over how much the VCA and mixer design in the Minimoog influenced that well-known sound. But I digress...

The first thing in the process of installing the Lintronics Minimoog modification is to remove the power supply board and check from broken solder joints and solder 'bridges' (some of which might be caused by resin still on the board that 'fouled' and started to conduct across filaments on the board). It sounds pretty mundane but is EXTREMELY important. As everyone knows, VCOs are sensitive to current and voltage changes, and even when using the SEM chips that 'feature' emerged early and often. So when you have shoddy manufacturing practices (the kind that lead to bad solder joints and voltage leaks across dirty/dusty contacts on the power supply) then it doesn't matter *how* awesome your audio design might be.

Let that be a lesson to everyone - this still applies today - especially on mass-produced computer audio interfaces that lay claim to stellar audio specs.

The second *big* thing that was done for the Lintronics mod was a 'piggy-back' Zilog Z-80 processor (on top of the Z-8, which was also replaced with a newer model that ran at a higher clock rate) for processing MIDI input and managing the interrupts into the primary processor for updating the various analog controls. Essentially it was turned into a 'dual-processor' analog hybrid keyboard. This meant that 1) controls didn't have to 'wait' as long for an update from the processor, and 2) there was less latency between changing a front panel knob (or MIDI control) and the resulting change in the sound.

What does this mean in the context of this thread? Well - many people who received their upgraded unit would call back and ask 'what did you do to make it sound better'? Aside from some re-calibration of the oscillators and going through the unit (using the original factory calibration spec) to make sure everything was ship-shape, not much. It's fairly easy to mistakenly dismiss their response as placebo effect. If you spend a few thousand bucks on a keyboard along with a complete retrofit of the instrument - the expectation is that is should RAWK - and rightfully so.



But there's that very subtle - and very *real* sense of 'feedback' that the musician gets from a properly responding instrument that makes the experience not only more gratifying for the player - but the musical result is unmistakable to the most casual listener.

My point - the quality of the *raw* audio sound of the Memorymoog had *not* changed, but the changes we made in the way the instrument behaved in response to input made a *huge* difference in the end result. I think that the point made about hardware synths and their low latency relative to running a plugin inside of a host environment has a factor that's much larger than the quality of the DAC.

[And now - I'm going to look like I'm contradicting myself by posting separately regarding a very recent anecdote that illustrates why the audio signal path matters]
Old 10th September 2012 | Show parent
  #45
Lives for gear
 
audslu's Avatar
 
🎧 5 years
Quote:
Originally Posted by subsonix ➡️
Yeah, I'd agree with most of this.

The same problem seems to appear in all the plugin synths I've tried.

Sure they have the general feel or frequency response to the hardware synth they are trying to emulate (or are influenced by), but there seems to be some kind of resolution factor missing from the plugins which make them soft sounding and cloudy in the mid-range.

It's almost like the coders are emulating the frequency response only and forgetting the dynamics.

It's in the dynamic of the attack that a lot of the magic lies with hardware synths and can make the difference between a tight punchy bass and a round mushy non-entity.

Not on an ADSR scale but something of a much higher resolution.

So I guess even on the digital hardware there is something happening in the output stage that adds some extra magic that coders are either missing, or that chews up so much cpu power that they don't bother with it.
This is from another thread, i hope bug doen't mind:

Quote:
If I remember correctly (math is getting a tiny bit rusty) the discrete part actually means, that the signal is supposed to be periodic (=infinite).
Obviously nobody cares, since some smart guy invented window functions.

Those tend to cause some tiny problems:
A window function tends to give you a wiggly frequency response (not as bad as it sounds, should be not really noticeable if you don´t window with a rectangle).
Also: windowing is very unhealthy for transients. This is why resynthesis tends to sound a bit mushy.
from this post:https://gearspace.com/board/8131443-post218.html


So i tend to believe in good code even more.

Have anyone noticed the difference between fm7 and fm8 (all fx off) or between versions of absynth until v4? Kontakt 2 vs. 4 ? ...this one (kontakt) i think improved in later versions while i prefer fm7 and absynth 1-3.

FM8 sounds to me more punchy but mushy and cold , fm7 is still one of my favourites, same with absynth more punch in v4 but...

And general i agree with transients being dull and plastic in some sw and i lately i noticed trilian has terrible attack very squashed (no fx) compared to electric bass library from another brand so it makes perfect sense to me know why i haven't used it so much since i bought it.

About fixed point vs floating point i have a poco pcie (24bit) and v-station sounds exactly the same with native, i can't hear any difference.

I believe it's a matter of algos, opamps and non-oversampling converters (then) vs oversampling ones (today).
Old 10th September 2012 | Show parent
  #46
Lives for gear
 
🎧 5 years
Quote:
Originally Posted by Michael E ➡️
Let me bring another metaphor.
If you are going to play the best new video games - you have to get a very good video card.
Without it everything will be emulated by CPU and result will be very pity.

Same with the sound.
Yes native stuff may be good, yes it does what it should, but... when you hear VA synth or hardware reverb - it sounds much better.
No, it's actually not the same with sound. Regarding your metaphor:

First: Video cards are a product for the mass market, that have been round for many years, so their evolution has been as fast if not faster than the evolution of modern desktop CPUs. That's not for DSPs which have always be niche market or where used for applications that are not so demanding on processing power (cell phones, mp3 en/decoder ect). Also all modern gaming engines are written for GPUs and are based on shader code mainly, while no one (except some nerdy demo programmers) writes CPU optimized gaming engines these days - but you would be astonished what highly optimized code would look like on a modern CPU.

Second: Modern 3D graphics rendering is much more demanding than audio processing. There's much more data to be processed in parallel and the requirements to the algorithms used are different to those in DSP processing.

And Third: While 3D specific functionality has been kept out of the CPU cores, because everyone owns at least a cheap GPU based video card, CPU's are capable of hardware accelerated DSP processing since 1997 (MMX) and have been continously improved since then (SSE1-SSE5).

So, yes, he was right, a modern i7 does outrun every DSP on the market by length, the difference in quality has other reasons (already mentioned here).
Old 10th September 2012 | Show parent
  #47
Lives for gear
 
🎧 5 years
Quote:
Originally Posted by bluegreengold ➡️
The other, probably more important factor, is that the software is written by people who didn't make the hardware, and that the software is it's own program and not the same as the hardware. Even when the same company makes a digital version for example, the korg legacy digital collection, they still make changes or add features to differentiate the software.
This.
Old 10th September 2012 | Show parent
  #48
Lives for gear
 
🎧 5 years
Quote:
Originally Posted by nznexus ➡️
i still think hardware sounds better then software...

but software are coming closer and closer.. but not yet close enough
You may be right generally (although some analog emulation plugs come close to the "hardware sound") - but not on this topic, because "digital hardware synths" are software running on a dedicated computer.
Old 10th September 2012 | Show parent
  #49
Gear Nut
 
HHaynes's Avatar
 
🎧 10 years
There's digital, there's analog, and then there's vintage

So - as to the sound of the analog path - for years I tended toward the 'placebo effect' camp. But recently I have shifted to the 'matters greatly in certain circumstances' camp.

It all started when I was hired by a fellow composer to do some basic audio re-work in his studio (setting up a patch bay) and training him on using Nuendo. He had a great setup to begin with, but he knew he wasn't really using everything to its fullest potential. So this was a bit of wood-shedding to make certain things more accessible *and* do some I/O tuning in Nuendo to make certain bits of hardware more accessible than it was when he was using Logic.

One of the things that he had installed was the Dangerous Audio 2-LT two-bus. It was connected to an Apogee Symphony I/O audio interface. The thing we wanted to do was put the two-bus in the Control Room path and also allow the returns to be used as an audio input when he wanted to record through the two-bus for stemming. It took some 'figuring' but essentially we were able to create 'Direct Out' connections in Nuendo that 'talk' directly to the stereo pairs that make the input into the bus. In the Control Room we set up an input monitor for the two connections coming back into the I/O. Aside from using it as a monitor source, it could also be used as a recording input for folding back from the 'analog bus'. It's pretty cool that Nuendo makes things that easy, but I thought we were spinning our wheels for what would be essentially 'placebo effect'.

Things became interesting when we started to go through the virtual synths that are mounted on the T.C. Electronics Powercore card - the Virus as well as the built-in synths that come with the card. I always start with the 'simplest' thing for testing, so I pulled up the instrument that looked like a Roland SH-101 (sorry, don't remember the name) selected the first preset and hit a key a few octaves below middle C. It made a heinous sound - I mean, a *great* sound from a digital synth perspective, but it was really big and gnarly. At that point I got the idea of trying to monitor the sound through the 2-bus so we pulled up the Control Room window and toggled between the 'in the box' mixer output and the return from the 2-LT.

The difference was stunning. Not mild - not subtle - like the placebo effect got smacked down on the playground and had its lunch money taken away.

The interesting thing about this situation is that it's the first time that *both* I and the studio owner had heard this. In many cases, when running broad program material (and if you're *always* monitoring through the bus) you don't necessarily hear what this 'custom vintage' analog signal path is doing to the sound. In this case, we have a low-end bass sound that's got a TON of high frequency content being poured into the bus on a steady basis. So when you see the meter showing little-to-no changes in peak and RMS levels, but you also see (and of course hear) the shift in distribution of energy across the audio spectrum, you can't deny that something very specific is happening. It bordered on magic.

Suddenly, all of those cranky raw built-in synths on the Poco card came out of the dust bin and are now viable (and great-sounding) voices worthy of adding to the mix. We then pulled up a newer hardware instrument - his Origin desktop synthesizer - and did the same thing. Right out of the gate, the Origin sounded pretty good when played back through the Apogee Symphony I/O. We found a sound that has some snarl in the bass - again to hear the changes that are made across the spectrum. We looked at each other and agreed that we didn't know *how* the 2-bus would improve on the warm, warbly sound. And yet, when we switched the monitoring the difference was immediately palpable - the effect was more subtle than the SH and Novation emulations on the Poco card - but definitely something we could hear.

So all of that time taking audio engineering classes with Moog - trying to understand how soft clipping in a circuit causes high order and low-order harmonic changes - suddenly makes sense in more than the context of creating another compression or distortion effect. I love my Arturia Origin keyboard - and up to now hadn't make much of a distinction when I run it via the stereo analog outs, the SPDIF out into my digital mixer, or the individual outs to the MOTU 2408mkIII (light-piping them into the RME audio interface for recording as separate instruments). But as of today I'm about 99% sure that the next major purchase for my studio will be a Dangerous Audio 2-LT. (and I'll probably either build or buy something similar for my live rig)

That's a very, very long way of saying that there's a LOT more to the sound quality equation than the make and manufacture of the DAC.
Old 10th September 2012 | Show parent
  #50
Deleted User
Guest
Quote:
Originally Posted by audslu ➡️
Have anyone noticed the difference between fm7 and fm8 (all fx off)...

FM8 sounds to me more punchy but mushy and cold , fm7 is still one of my favourites...
I don't know what you want to say, but I definitely agree that FM7 sounded better than FM8. FM8 was soundwise a great disappointment to me as successor to FM7.
Old 10th September 2012 | Show parent
  #51
Lives for gear
 
🎧 5 years
Quote:
Originally Posted by Eric J ➡️
No. I'm saying a DSP chip in a digital synth can dedicate 100% of it's power to making sound while your computer CPU cannot (since it has to eat up cycles to perform other tasks).
the computer cpu is often far more powerful than middleage dsp chips.. but the development time yamaha has spend is far deeper than a liitle coding form a rather small software company..

beside that hardware synths have analog outputsatges that add to the sound and would need to be moddled with the same effort than the best vintage modeling plugs.. and as we know theese vintage modeling plug still dont nail that kind of subtile colorations.. and last but not least.. low latency operation also effects performance and therefor quality of the outcome--

analog processing on digital hardware synths also achieves a great deal on beefing them up

all this hit records featuring an elektron machine drum for example.. thats not the sound you would get by directly AD the output of the MD..usually things labled manley or similar are involved...
Old 10th September 2012
  #52
Lives for gear
 
🎧 5 years
Random thoughts:

Voodoo in the output stage:
Probably true for some older stuff. You probably want to do a DC cut and you need a final output stage anyway. With a couple of well placed capacitors this is already nearly an EQ. (well a very bad and simple one).

This can explain some "openness" and "thinness" (high DC cur) and a "warm character" (6db lowpass at something like 8kHz).

Would not count on this with anything modern. Most users would be unhappy, if their digital outs sound very different from the analog outs. PC3 sounds quite similar on digital and analog outs and I guess it´s a quite nice sounding synth.

So, sound difference for the modern stuff is all in the Calculation.
I know, that a lot of people will ask, why not build great sounding algorithms into softsynths. I think they exist. A lot of Softsynths sound excellent in their own way. And I´m not only talking about modern CPU hogs.

So differences:
FX: Yes, I know, real synthesists don´t use them and that is very reasonable on say a Blofeld, but for anything remotely workstationlike (including e.g. the Virus and the Fs1r) you normally get a set of quite decent FX. Software less so.

I recently experimented a bit with running Absynth through some of the guitar rig stuff, just a cabinet (no Amp), some mic placement a tiny bit of the simple studio reverb thingie, no Abynth "I am so spacy" FX. Suddenly I had a Synth with a very different character. Same goes for EQing etc.

So, if in doubt switch of every synth FX and add your favourite other stuff.

Other stuff is less easy to fix. Let´s start simple: everybody cheats in digital synthesis (well, exceptions (or close to exceptions) are the Solaris, Aalto, maybe Diva, probably Bazille, some Reaktor stuff)).

Basically running everything at audiorate will eat CPUs (even modern ones) for breakfast. So you have a bunch of slower modulation. Obviously different people cheat in different places.

Another major difference is Envelope shape. I think basically nobody really uses linear or exponential. There is always a slight difference to the perfect world, or compromises to do nice Amp and Pitch mod or something. Obviously you can shape your own Envelopes with many modern Softsynths, but be honest, how much time to you normally spend to shape your attack?

One last thing, before I quit, and this might be the decisive one:
A bunch of softsynths have only limited modulation by Midi controllers, especially velocity and note number. Even when they have it, many of them will throw a default sound at you that has no modulations assigned.

The Fs1r for example is the polar opposite. I think every Operator has several points where Velocity and Keytracking can influence it. So your sound will feel a lot more "alive" when you are playing. PC3 (and other Kurzs) seem to be similar. I think most stuff has hardwired velocity control.
Old 10th September 2012 | Show parent
  #53
Lives for gear
 
🎧 5 years
Quote:
Originally Posted by HHaynes ➡️
So all of that time taking audio engineering classes with Moog - trying to understand how soft clipping in a circuit causes high order and low-order harmonic changes - suddenly makes sense in more than the context of creating another compression or distortion effect. I love my Arturia Origin keyboard - and up to now hadn't make much of a distinction when I run it via the stereo analog outs, the SPDIF out into my digital mixer, or the individual outs to the MOTU 2408mkIII (light-piping them into the RME audio interface for recording as separate instruments). But as of today I'm about 99% sure that the next major purchase for my studio will be a Dangerous Audio 2-LT. (and I'll probably either build or buy something similar for my live rig)

That's a very, very long way of saying that there's a LOT more to the sound quality equation than the make and manufacture of the DAC.
I totally believe you - I recently racked two channels of an old D&R Series 8000 mixing desk, equipped it with Lundahl output transformers to use the EQs and preamps as standalone units into my Layla 3G interface. The first time after hours of soldering i plugged one of the channel strips in and ran some tests with Mics and DI guitar, I was totally overwhelmed by this rich, warm and big sound I suddenly got, and when I switched on the EQ section, I almost started to cry (literally ). I have lot's of EQ plugins, and was quite content with them up to the day, but none of them is capable of doing that to a signal. I already had this kind of "anti placebo" experience when going from interface stock pres to an Art Pro MPA II tube preamp, but the D&R strips topped even that by far - not only in overall sonic quality - the D&R preamps don't even produce a whisper of noise at max level while the MPA II stays only noiseless up to two thirds of the maximum gain/output level on most sources.
Old 10th September 2012 | Show parent
  #54
Lives for gear
 
🎧 5 years
Quote:
Originally Posted by bug2342 ➡️
Random thoughts:

Voodoo in the output stage:
there is no vodoo..just the sound of the used components and quality of powersuply that gives an hardware synth stage in most cases more power than you have on an average soundcard,,,easy to get when your synth gives much more output than your soundcard can take..

and any opamp has its own sound.. the warm sound of many japanese synths is more based on the 4558 stye op amps than on a special dsp algo
Old 10th September 2012 | Show parent
  #55
Lives for gear
 
rafkey's Avatar
 
🎧 10 years
Quote:
Originally Posted by Michael E ➡️
- Powercore project seems just dying.
- UAD has no synths.

Same with the sound.
Yes native stuff may be good, yes it does what it should, but... when you hear VA synth or hardware reverb - it sounds much better.
Yes! the ears rule
Old 10th September 2012
  #56
Lives for gear
 
kpatz's Avatar
 
🎧 10 years
I have a thought or two on this...

Years ago, when you had a computer that displayed graphics on a monitor with a 640x480 maximum resolution, and displayed a photo that was in 640x480, it looked pretty good. Certainly not as good as today's multi-megapixel images, but good nonetheless. Now, with your modern computer with high-definition monitor and graphics card, open up that same 640x480 image and view it full screen. It doesn't look so good anymore, does it? Upconverting the image to a higher resolution either makes the image blurry or pixelated, depending on the algorithm used to do the upconversion. The resolution of the hardware has improved, but the source material (i.e. samples used in a software emulation of a sampler) hasn't.

Also, for soft synths, the algorithms used to emulate oscillators, filters etc. may be optimized for a certain sample rate/quality that is lower than the output of modern sound cards/interfaces, in order to limit CPU load to allow more polyphony, instances, or for running on older hardware. Even though your Core i7 CPU is far more powerful than the chip in a Nord Lead 2X, the soft synth designer can't use all that power since it wouldn't coexist very well with your other software (DAW, etc.) The designer of the Nord Lead 2X can use all the CPU or DSP power to generate sound.

Also, the development cycle for a soft synth is likely to be a year at least, and hardware evolves quickly. The newly released "SuperDuperSoftwareSynth 1.0" was probably written to run on a 3 or 4 year old system. It won't necessarily take advantage of the faster CPU speeds of a newer system except perhaps to allow more polyphony, effects, or instances to be run at once. But since the native resolution of the soft synth doesn't always match the native resolution of the sound card, a conversion has to happen, and like when you blow up a 640x480 image on a high resolution monitor, it just isn't going to sound the same.
Old 10th September 2012 | Show parent
  #57
Lives for gear
 
🎧 15 years
Quote:
Originally Posted by audslu ➡️
This is from another thread, i hope bug doen't mind:
The thing about the window function makes a lot of sense to me intuitively.
Old 10th September 2012 | Show parent
  #58
Gear Guru
 
Yoozer's Avatar
 
1 Review written
🎧 10 years
Quote:
Originally Posted by Michael E ➡️
- Powercore project seems just dying.
- UAD has no synths.
Compressors, EQs, et al are instruments you stuff in a rack without a twinge of remorse. Synths, not so fast. Clavia tried it with the G2 - USB for the editor, and no knobs on the rack. No thanks; if it dies you have a paperweight that doesn't even look good.

Creamware still offers external DSP. Kyma as well.

Still, a pure USB stick would mean that you don't get the analog outputs, and that's a factor.

Quote:
Let me bring another metaphor.
If you are going to play the best new video games - you have to get a very good video card.
Without it everything will be emulated by CPU and result will be very pity.

Same with the sound.
Yes native stuff may be good, yes it does what it should, but... when you hear VA synth or hardware reverb - it sounds much better.
No, not the same with sound! Here's why your argument is bogus, and why everyone arguing for "100%!" should first understand the basics of a Turing machine.

You can calculate 2+2 by hand, on a sheet of paper.
You can calculate it by counting rocks, or apples, or whatever.
You can leave it to a calculator.
Or, you can tell a computer to do it.

In all cases, the outcome is 4. It does not matter how long it took you to sum those numbers, the outcome is the same.

The reason that a game looks like **** when it's running on the CPU instead of the GPU is not because the GPU does anything special. It does not use 2s instead of 0s and 1s. Each calculation, transform, and blend can be done on the CPU without any issue.

What the CPU can not guarantee is the same framerate. So, in order to keep the game playable at 30fps, the number of calculations is reduced, which means that the more complex stuff - shading, textures, lighting and detail - is reduced.

Quote:
Originally Posted by zerocrossing ➡️
So you can have an amazing DAC on a digital synth... but doesn't it eventually end up going into your audio interface anyway?
Yeah, that's the weird part. You'd think those expensive DACs would be any good, but turns out they're just neutral. So - softsynth through neutral DAC = crap?

Quote:
I think the actual truth, that's already been mentioned, is that a good digital synth uses 100% of it's power for a singular purpose with no compromise.
That's nice but no.

The 5% an i7 spends on the algorithm are still more instructions per second than an old DSP. Even if the DSP needs 2 clock cycles for a MUL or LSH and the CPU needs 40, there's power to spare.

As shown, there's a lot of compromise in a VA. Anything at non-audio rate in the Nord Modular doesn't get the full 96k treatment. Aliasing in various synths (no oversampling), including the Virus, JP8000, etc. Polyphony drop (micro Q) when more modulations are enabled.

Quote:
Originally Posted by HHaynes ➡️
The FX are part of 'the instrument' and therefore is an inextricable part of the sound. Ask any DSP coder (after a drink or three) and they will tell you all about 'the gives and gets' of generalized versus dedicated platforms.

I have more to say/type about this but will switch to my desktop after my morning coffee.
I thought your story was great (though it's Curtis Electro Music, not Synthesizer Expander Module). Also, you're in the minority with saying FX are part of the sound; most folks here grudgingly accept the chorus on the Juno as unavoidable but would rather have that Access spends more time on the oscillators and less on the FX.

(though I really really like the FX in the Virus - great phaser, great distortions)
Old 10th September 2012 | Show parent
  #59
Lives for gear
 
🎧 5 years
Quote:
Originally Posted by living sounds ➡️
The thing about the window function makes a lot of sense to me intuitively.
Will since this was dug out:
Yes, Window functions tend to kill transients in samples.

The bad news (well actually it´s good news, but bad for instant thread gratification):

Very few softsynths use window functions.
I don´t know the code of the VSTs, but windowing basically means your sample replay will eat approximately twice as many CPU cycles and you will actually not get a gain when running some periodic waveform through it.

[Off topic: the above is not strictly true, you can do formant shift stuff etc. See e.g. Virus Graintable/Formanttable]

Also: most VSTs create their transients after the Oscillator/sample play section. Those would not be bothered.

Bad Envelopes are way more serious offenders. Especially since a bunch of the popular plugins today still drag some baggage from the time when things where called CPU hog for doing things right.

Random example: The otherwise beautiful Reaktor still defaults to 400Hz control rate and maxes out at 3.2kHz. Some Waldorf LFOs go to twice that frequency....
Old 10th September 2012 | Show parent
  #60
Lives for gear
 
audslu's Avatar
 
🎧 5 years
Quote:
Originally Posted by Analog Prophet ➡️
I don't know what you want to say, but I definitely agree that FM7 sounded better than FM8. FM8 was soundwise a great disappointment to me as successor to FM7.
I mean what you agree with.
📝 Reply

Similar Threads

Thread / Thread Starter Replies / Views Last Post
replies: 627 views: 104769
Avatar for damoog
damoog 27th April 2019
replies: 835 views: 177322
Avatar for jlgrimes11
jlgrimes11 30th August 2020
replies: 518 views: 41619
Avatar for audslu
audslu 28th December 2015
replies: 498 views: 60861
Avatar for GlenGood
GlenGood 4 weeks ago
Topic:
Post Reply

Welcome to the Gearspace Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…

Forum Jump
Forum Jump