Location: Somewhere between France and Absurdistan
Posts: 1,466
Thread Starter
Synclavier emulation ?
Calling all plugin developpers : Would you please make an AU/VST emulation of the FM section of the Synclavier 2 ? Pretty please ? I BEG YOU ?
In fact could you just emulate the whole thing including the sampling section ( yeah i know 100 KHZ converters and all that ), if not, just the FM section would be already great..
Not that i expect this to ever happen..but who knows..
For the kiddies who don't know what a Synclavier sounds like , here are some songs from an 1985 album by Eddie Jobson, done almost 100% on a Synclavier ( and one of my favorite albums ):
__________________ "Most people are fools, most authority is malignant, God does not exist, and everything is wrong." Ted Nelson " Imagination is a quality given a man to compensate him for what he is not, and a sense of humour was provided to console him for what he is." Oscar Wilde "99% of musicians are not business people, which is why everyone makes money off of them - except them." Sqye
wanted to imbed a youtube video for the synclav II fm...ohh well.
I have had a couple of hands on sessions with a massive Synclavier system that was owned by a Hayden Clemet at post logic studios in LA.It had a 16 track direct to disk and 32 stereo voices and an amazing poly pressure keyboard.Anything you could think up...you could do it.It was often used for film and tv and its massive string library was constantly in use.Sometimes it was used to demo a film score before booking a real orchestra....or sometimes it work in tandem with a real orchestra to phatten the whole thing up......or simply put...IT WAS THE ORCHESTRA.I saw Eddie Jobson passing through the hallway a few times there.These systems are still in use today but as most know...times move on.
I am working on Synclavier emulation on the Kurzweil PC3, using a BCR 2000 to control it (specifically because its knobs can do 14 bit high-res control with four levels of programmable speed response). Video will be available soon-ish.
The problem is that variable sample rate systems are difficult to emulate, since it's a given that software is going to be fixed sample rate. The zero order hold reconstruction (i.e. nearest neighbor interpolation) adds high harmonics (image frequencies), but these never alias. This is the "classic" sound of the Fairlight, Synclavier, Emulator, etc. that you can't get with modern machines. This is especially a problem with the FM voices, since the waveforms are very low resolution, and a lot of the Synclavier's signature sounds seem to use a very high modulation index. You wouldn't want to get so crazy on a DX7 since there will be too much aliasing.
I'm working on a way to add band limited image frequencies without oversampling (not with a product in mind, but maybe a paper or something). But FM will still need lots of oversampling.
Great tracks. I have been experimenting with these more 'widescreen' digital soundscapes and have found so far later DX synths very useful. You might find a DX7II or a TX802 a useful addition to your set up while you wait for the ultimate synclavier replacement. I was very pleased with the 'synclavier' / 'fairlight' and PPG esque sounds - on the DX7 to - epspecially layerd with analog textures.
The problem is that variable sample rate systems are difficult to emulate, since it's a given that software is going to be fixed sample rate. The zero order hold reconstruction (i.e. nearest neighbor interpolation) adds high harmonics (image frequencies), but these never alias. This is the "classic" sound of the Fairlight, Synclavier, Emulator, etc. that you can't get with modern machines. This is especially a problem with the FM voices, since the waveforms are very low resolution, and a lot of the Synclavier's signature sounds seem to use a very high modulation index. You wouldn't want to get so crazy on a DX7 since there will be too much aliasing.
I'm working on a way to add band limited image frequencies without oversampling (not with a product in mind, but maybe a paper or something). But FM will still need lots of oversampling.
I'm of course not familiar with synclavier's fm engine, nor hardware design, but I understand what variable sample rate is and implies. I wonder thus how to relate its advantages to even the simplest form of fm (one modulator/carrier pair), unless their frequency relationship is limited to simple integer ratios, or the hardware relies on some intricately hybrid analog/digital technology
Location: Somewhere between France and Absurdistan
Posts: 1,466
Thread Starter
Quote:
Originally Posted by zerocrossing
Good idea posting this on a mostly hardware orientated website.
?
Sorry, i'm not religious. ( as in, not interested in hardware vs software, analog vs digital, mac vs pc, vintage vs modern religion wars ). Only interested in music , yeah i know, that's weird.
?
Sorry, i'm not religious. ( as in, not interested in hardware vs software, analog vs digital, mac vs pc, vintage vs modern religion wars ). Only interested in music , yeah i know, that's weird.
What does religion have to do with posting a call for software developers in a mostly hardware orientated forum? All I'm saying is that you'd have better luck finding a software developer on a site like kvraudio.com than you would here.
I have the 'Synclav FM' sample CD-ROM for the Kurzweil. Frankly, it was a mostly disappointing purchase. A few sounds are interesting, but I find that you could obtain better, cleaner FM sounds with a DXII or a SY99.
Of course, the fact itself that they're sampled could be a big factor in lessening their appeal...
Are you kidding me ? Is there an unwritten law on Gearslutz that forbids talking about software or plugins ?
Noone is trying to give you a hard time. Just the opposite. There is software talk here, but almost always in relation to hardware. It is *gear*slutz after all. 95% of the threads are hardware related. Suggesting that you post to a more software/ computer programming focused forum is just trying to help.
I'm of course not familiar with synclavier's fm engine, nor hardware design, but I understand what variable sample rate is and implies. I wonder thus how to relate its advantages to even the simplest form of fm (one modulator/carrier pair), unless their frequency relationship is limited to simple integer ratios, or the hardware relies on some intricately hybrid analog/digital technology
I think it's not really that complicated. It just ended up being a sort of stupid design because Cameron Jones couldn't think of a more clever way to implement it, and by coincidence it ended up being free of objectionable aliasing.
I'm not entirely sure about the modulator/carrier interaction. The modulator and carrier seem to have their own independent readout rates, and from there I can think of two possibilities: either the phase adder, wave ROM and DAC respond instantaneously to changes in both the modulator's output and carrier's phase, or the final output rate is synchronized to the carrier's readout rate. The latter makes more sense, because you generally want a sample and hold after the DAC to avoid glitches on transitions... but this would imply that the modulator is "resampled" to the carrier's rate. I don't think it would adversely affect sound quality much because it's just a sine wave, but it might not work as well for non-sinusoidal modulators, operator chains, etc.
I've been wondering lately if it would be possible to do a true variable sample rate quasi-emulation in hardware using a cheap microprocessor and DAC to generate a single voice. That's basically how all the Wersi stuff works, but microprocessors are a lot cheaper and a lot more powerful now. I think it would offer some novel possibilities that are generally a bad idea in software, i.e. extreme waveshaping and FM, low level bit fiddling, etc.
Are you kidding me ? Is there an unwritten law on Gearslutz that forbids talking about software or plugins ?
Pay no attention to that guy. This is where I also read both software and hardware threads. Software discussions here are just as valid as that guy continuing to type "orientated" instead of "oriented".
The enhaus link mentioned earlier has a few nice things. I just downloaded the free sounds. Has anyone bought the full, bigger version of the enhaus vsti?
I used a Synclavier here In LA in 85 or so. Back then, I couldn't afford one other than the rentals. I ended up sampling that popular old blue see-thru vinyl demo that NED had back then. Not quite like the real thing, but it was something.
__________________ "make multitrack sound for long long time" "I don't understand this shootout. May I borrow your ear canals so that we're on the same page?" "Lofi is an artform....not a sample rate""
Just to be clear, orientated is a bona fide British English word. If you won't pick up on that, we Brits won't pick up on the US misspelling of words like 'plow' or referring to the bonnet of the car as a hood...!
Anyway, this is electronic music instruments, not just hardware, so I'm in agreement that anything relevant goes here. The GS forum reaches out to a slightly different demographic than KVR and I guess a thread across both is the internet's best way of finding the definitive answer..
LOL people actually trying to recreate that? Good luck. Nothing even comes close to the original. Your living in La La land... It seems to be normal around here lately though, so at least you fit in. (thumbs up).
LOL people actually trying to recreate that? Good luck. Nothing even comes close to the original. Your living in La La land... It seems to be normal around here lately though, so at least you fit in. (thumbs up).
I remember there was a patch emulation for the Nord Modular G2 (I think the creator of that patch is a member here actually, not sure about his name...).
I tried it and it was pretty cool indeed.
But a real Synclavier with the VPK would be really really great to have for sure
Location: Somewhere between France and Absurdistan
Posts: 1,466
Thread Starter
What's up with the sudden "Gearslutz is a 99% hardware forum, i suggest you post elsewhere for software" complete nonsense ????
Do you guys ever venture outside the "Wich mono analog vintage synth should i buy" threads ?
Take a look at other sections of Gearslutz, you might be shocked. HALF the threads on GS are about software. There is even a songwriting subforum now !
In my 9 years on this forum , I've never seen anyone going into those endless Protools or Waves plugins threads to suggest "Hey , maybe you should move those Protools conversations over to KVR , cause you know, this a mostly hardware oriented forum.."
I'm just trying to start a discussion about a HARDWARE DIGITAL synth and post-production system called Synclavier, see what other people think about it, their experiences, what software alternatives or emulation might be possible , etc...
It's not like i'm doing some official call to hire C++ developpers or something like that. It's just wishful dreaming about an FM Synclavier in a software form i'd love to have...
Anyway, sorry about this long rant, but this is just so beyond weird and clueless...
Back to subject.
Interesting opinions and suggestions so far, please keep them coming...
It's not like i'm doing some official call to hire C++ developers or something like that.
Even then, going to KVR would net you more reactions like acreil's (actual in-depth technical information on why emulation would or would not be possible, and how hard it would be) and fewer like alexp's, simply because there are more developers there than there are here.
You might even encourage someone to just give it a try to create a single variable sample rate voice as a quick sketch in SynthEdit or whatever, or with some C/C++ code that's still lying around.
So with the variable rate there are two things: all voices are completely independent (meaning that the sampler does not have to worry about synchronizing things with a common multiplication factor or something), and the control over the playback frequency is very accurate (measurable). Furthermore, the DACs impart a character on the sound. These are all things you have to model.
I think dealing with differing sample rates is already a solved issue (for low rates) in the video editing industry, since you're dealing with 24 fps, 25 fps, 29.97 fps and all kind of crap and it has to be mixed properly (if you crossfade two videos). But it might very well be that there it's solved by converting/resampling all videos to the same frame rate (or an internal theoretical framerate that's high enough to hide most artifacts).
Thing is, you're dealing with higher speeds here (100 kHz) and of course you want sufficient polyphony.
I'm kind of surprised that nobody's tried to resurrect this in hardware. It could very well be that it already exists in some Eurorack module - (MakeNoise Phonogene?), but then it's limited to 1 voice and the rest of the structure's not there yet.
On the other hand, equivalents of most of whatever was in the Synclavier should be really cheap nowadays, thanks to Moore's Law and all that.
__________________ For all the intelligence and knowledge that technology empowers us with, the lazy and stupid is amplified along with it (Staticstarter) Threads to check out:Chord Generators & Tips | Pop Sound Sources
Location: Somewhere between France and Absurdistan
Posts: 1,466
Thread Starter
Quote:
Originally Posted by Yoozer
I'm kind of surprised that nobody's tried to resurrect this in hardware.
Yep. Maybe it's too complex after all to recreate in a plugin version, but i'm sure there's a market if they resurrect it in a new hardware synth version for a much cheaper price, and it could be quite successful.
I wonder if the owners of the brand name are still active...
Didn't the original Fairlight makers actually resurrected the whole 2X model recently, under a limited edition ( and for a crazy price..) ?
Yep. Maybe it's too complex after all to recreate in a plugin version, but i'm sure there's a market if they resurrect it in a new hardware synth version for a much cheaper price, and it could be quite successful.
I wonder if the owners of the brand name are still active...
Didn't the original Fairlight makers actually resurrected the whole 2X model recently, under a limited edition ( and for a crazy price..) ?
Yes, Peter Vogel was/is putting out a Fairlight CMI 30... not sure if it happened though.
I'm kind of surprised that nobody's tried to resurrect this in hardware...
Looks like both the Synclavier II and the 9600 are being built/restored/and sold via Synclavier's Europe site. When I was using a couple here in LA in the early 80s, NED wanted $63,000 for one of them. No idea what Synclavier is selling the units on their current site for. Should maybe send them an email to find out.
I was just browsing the current site a few minutes ago. Tons of stuff there. They mention a future plan for more extensive 3rd party sound libraries. But from the looks of the existing hardware inventory, they're selling a fair amount of hardware systems. Maybe enough to not make them very anxious to develop anything 3rd party.
Also, as others have mentioned, there's a lot to constructing a sound on the Synclavier
I'm trying to think of another way to do it that would do the FM with no aliasing, going with the assumption that the DAC is "synchronous" and doesn't instantaneously change according to the phase of the modulator...
Good quality band limited interpolation would permit 20 kHz bandwidth at almost any playback frequency.
If I have time I'll try to make a rough prototype in Octave.
For under $5k you can get into a fully refurbished Synclavier - all the way up to $16k for the big daddy - and all of these systems have the fantastic VPK keyboard, which is a big part of the experience of playing the beast. A buddy of mine has worked with these guys to get his system upgraded and they really know what's up and have extensive spares and systems in stock. He bought a system for a couple hundred bucks at an electronics salvage place here in LA and these guys got him up to speed for pennies on the dollar...
I did a quick emulation attempt as a proof of concept/educated guess. I'm using 256 point tables for both the modulator and carrier (which I think is right). I made two versions to cover both synchronous and asynchronous cases that i mentioned above. The synchronous one seems less glitchy, but they sound pretty similar.
I'm a little disappointed at the fizzy noise, though it seems to be pretty much in line with how the Synclavier actually sounds (not that I have one to compare it to). It's due to the resolution of the lookup tables. I should be able to make a workaround that will make the FM smooth as the modulation index changes, but keep everything else properly nasty.. but it won't be authentic.
The sound is inharmonic FM. The carrier has harmonics 1 and 5 and the modulator/carrier ratio is 3.13. It's just the first thing I thought of, so it's not exactly brilliant sound design. Also it's obviously just one voice.
New Reply 
Facebook
Twitter
LinkedIn
Good idea posting this on a mostly hardware orientated website.
