Another explanation that I've heard is that R&D for synthesizers in the 1980s and 1990s got brutally expensive, and a lot of companies just didn't make the transition.
Building synthesizers out of discrete components was simply far more expensive than using digital technology to create and modulate various sound waveforms. The Yamaha DX was far cheaper to manufacture than the polyphonics Prophet 5, Oberheim OB-Xa, or monophonic Minimoog. The DX was lighter and more reliable, produced a wide array of sounds, and was easier to keep in tune despite being more difficult to program for useful sounds. Many DX users used the stock sound patches except for Brian Eno and a handful of others who could dive into FM synthesis.
The Minimoog and Prophet 5 are both from the 1970s. Despite the Minimoog's status as a collector's item today, it had roughly the same retail price as the DX7 in the 1980s--around $2000 USD. The Prophet 5 was more expensive at $4000, but it was also an incredibly successful, flagship synthesizer.
Of the synthesizers listed--Prophet 5, Oberheim OB-Xa, and Minimoog, only the Minimoog was made primarily out of discrete components. The others relied heavily on integrated circuits from SSM or CEM. (Obviously they all contain discrete components--but the synth modules in a Prophet 5 or Oberheim OB-Xa are basically ICs with some support circuitry.)
Sequential went out of business after developing the Prophet VS. Although it bears the "Prophet" name, it is a digital (or hybrid) synthesizer. It uses wavetables (like the PPG), but four of them simultaneously, with a Curtis VCF. It had a novel architecture (therefore confusing), the synthesizer itself was expensive, and there were some QA issues. It used TMS DSPs plus some custom ICs designed by Sequential. Apparently, during the design of the Prophet VS, there were various problems with the ICs that were fixed with kludges--Sequential wasn't in a financial position to make revisions to the ICs.
In other words, the R&D cost for digital synths was too high.
By the mid-80s the analogs were seen as last decade's sound for last year's people. They were immensely unfashionable, and by the beginning of the 90s they were being sold off for bargain prices.
The low end of the analog market had been captured by budget polysynths from Korg and especially Roland which had most of the sound at much less of the price.
The problem wasn't the cost of digital R&D, it was commoditisation which turned a high-margin super expensive boutique professional market into a low-margin semi-pro and amateur commodity market with cheaper, less exclusive and less prestigious products.
The high-end market essentially disappeared and so do its distribution and promo channels. What was left supported a much smaller number of companies.
All of the headline synth companies went under at that point, no matter what tech they used - Sequential, Oberheim, Moog, PPG, ARP, Ensoniq, E-mu, and even Fairlight and eventually Synclavier. As well as also-rans like Octave, Gleeman, Crumar and Siel.
Korg also went under and were bought out by Yamaha.
Yamaha survived because it's a huge zaibatsu with deep pockets. Even with the DX7 the synth division is a sidenote on its bottom line.
Roland are also a huge corporation - mostly immensely profitable while also having street-level cool - and weathered the 80s and early 90s with some functional but clever hybrid models at relatively affordable prices.
Around 2005 the boutique market started to reappear, but aimed less at professionals and more at now-affluent collectors and dabblers who "always wanted one of those", and/or are willing to drop $$$$$ on something spendy just to prove they can.
So you get hipster-toy module makers in the Eurorack space, wildly overpriced Veblen toys like the OP 1, remakes of Classic Brands™, and various clones and reissues in both hardware and software form.
It's very nostalgic and backward looking, and I suspect there will be another shakeout before the end of the decade.
> So you get hipster-toy module makers in the Eurorack space
Which are used by artists all around (from amateur hobbyists to professional acts), they aren't "toy modules", there's been a lot of innovation in Eurorack and painting all of them as "toys" is a huge disservice to the whole format...
What ended up working during the late 80s and through the 90s were ASICs that could be reused in areas other than just "pure" synthesizers - digital pianos, home organs, PC soundcards (IIRC both Emu and Ensoniq ended up being bought by Creative, the SoundBlaster company).
So this worked out well for those companies that were in a position to do a good job of both the ASIC and logic board development, and also use the resulting technology to address those broader markers - Yamaha being the pre-eminant example. (I agree with the point about the DX7 being a sidenote, but I'd say that all the other product lines that the tech eventually flowed into will have ended up being rather more than just a sidenote.)
Regarding the current market, despite all the retreads, I do think there's more interesting stuff going on than there has been for a long time (certainly in comparison to the analogue synth revival of the 90s), and I hope that manages to continue happening, even as I also kind of hope that the retro-appeal stuff will simmer down a bit.
I am a fan of the analog renaissance! They sound great and are fun to play. It's not just classic reissues (and improved versions) or clones either - there's a huge range of analog from cheap pocket synths to monster synths to hybrid to modular.
One particularly great development is more affordable analog from the IK Multimedia UNO to the Korg Minilogue to the Sequential Take 5 to Behringer's Deepmind and Neutron (not to mention their clone army if it ever reaches production.)
"The others relied heavily on integrated circuits from SSM or CEM. (Obviously they all contain discrete components--but the synth modules in a Prophet 5 or Oberheim OB-Xa are basically ICs with some support circuitry.)"
I am currently recapping/restoring an Oberheim OB-Xa 8-voice. I just finished the power supply and now have eight voice cards that need about 13 parts, capacitors and mini-potentiometers, that need to be replaced on each. Each voice card has thirteen ICs but over one hundred discrete components. The main board has dozens of ICs and many other discrete components. The ICs replace the discrete components that comprised the oscillators, filters and VCAs but there are over a thousand discrete components in this synth.
Comparing a Minimoog with a DX7 is apples and oranges. They were released 13 years apart. the Moog has one voice while the DX sixteen voices. The U.S. inflation calculator puts the Minimoog at $11,884 in today's dollars and the DX at $5,790.
The VS does not use wavetables like the PPG. Vector synthesis is in a category of its own. It uses four single-cycle waveforms that can be selected individually, and one morphs among them in a 2D plane using a joystick or via envelopes and LFOs. It is not a crossfade, as it recalculates the resulting waveform on the fly. The result is similar to a wavetable, but in 2 dimensions.
Well I'm pretty sure the oscillator bank in the VS does use wavetables, it would have no other way of supporting 127 different voices (including user-provided voices). (But you're correct that the approach to modulating sounds on a VS is fundamentally different to, say, a PPG wavetable synth.)
That's not what a wavetable is, by any definition relevant to synthesis.
Of course the VS has waveforms stored consecutively in ROM, with 32 more stored in NVRAM. That's just data. In wavetable synthesis, an entire table of consecutive waveforms is loaded into an oscillator. In vector synthesis, 4 individual and arbitrary waveforms are loaded into an oscillator.
I think of a wavetable as a system with multiple short, periodic waveforms, where you can interpolate between them. On the PPG, you get something like 32 waveforms, and interpolte between consecutive ones. On the Prophet VS, you get 4, and you interpolate between all 4 at the same time.
It sounds like you use a different definition, which is fine. I don't think we're going to learn anything by litigating this kind of disagreement.
Yes and cheaper still is to just use the computer you already have to be your synth. I don't think there are any advantages today of using an analogue synth. Software synths can not only do all those "fat" sounds but can do a really good job of synthesizing real instruments such as an acoustic piano. (see Pianoteq, for instance, which does not use sampling to create instrument sounds)
I find analogue synths pretty fascinating but if you are just using them to make music (as opposed to a hobby in themselves), there are far better ways, here in 2022.
Not to take anything away from Dave Smith, he was brilliant and his synths were the best you could get, in their day.
This is mostly true for the sound of the instruments – the emulations are very good these days – the folks writing them often sit with the original instrument and an oscilloscope and adjust the software until the outputs are the same.
But there are two things primarily that you get with physical instruments:
- A bunch of limitations. You're more likely to dive into deeply understanding your instrument if it can only do a well-defined set of things. This is counterintuitive, but absolutely true. If you're using VSTs, you should similarly force yourself to use only a small set of them, and to swap them out very rarely.
- A tactile interface. Generic MIDI controllers just don't do as well. There is a gap in playability between the two.
That said, I usually explore on a physical synth and often end up using a soft-synth in finished tracks. A physical synth is a lot more fun to jam on than a MIDI controller. It may be possible to build a physical controllers or modular controllers that captures more of that feeling (and a lot of high-end physical synths these days are just ARM systems running real-time Linux), but the run-of-the-mill MIDI controllers are not that.
(Reference: have worked for two companies that produce VSTs, including the biggest of them)
Exactly this. Physical synths are more fun to play with, explore, learn. You can feel you fully understand them (I challenge anybody to say they fully understand every feature of every module of native instruments and arturia v collection, which together cost as much as a single mid range hardware synth :-). They are great for hobby and creativity, playing around and having fun, and in some ways for performance. They are tactile and immediate and physical. (Note that I am empathically not claiming "superior sound" in that list ;)
Daws are practical for actually making and producing music.
In reality though they can complement each other nicely. You can sketch on hardware and even integrate it into a daw.
----
My personal thought on superiority of sound is same as records and film - if you feel a particular specific past sound (or image) is the pinnacle of possibilities, soft synths will indeed asymptotically recreate it and may or may not ever recapture whatever "feel" it is that makes that specific sound superior to you.
I came into this late and without preconception so all sounds are equally valid with no absolute framework and all synths CAN produce interesting cool useful and usable awesome sounds - I'd rather marvel at creativity and variety than endlessly try to recapture something previous (which even my favourite companies are sometimes stuck doing - how many ways can roland and others re re re re re create the Jupiter 8 sound?? Same for Moog etc)
VR synths have just recently started to come out. A VR interface to VCV Rack would convince me to buy a VR rig. The link below is to Synthspace, a VR modular kit, and that’s got me on the fence already.
You still miss out on knobs/buttons/switches etc., but you can make use of the spatial memory that each synth/module’s design promotes, and that’s pretty big. I can envision a VR controller extension with a physical knob for adjustments that would improve the experience, but it’s probably too niche a product to be built anytime soon.
https://sound-force.nl/ make software-specific midi controllers. They provide knob-per-function control of cool software synths.
Interestingly the softsynths they target are clones of classic synths. It gives people the classic interfaces, but with a lot more flexibility with tracking and automation. They seem cool.
Physically modeling a Rhodes somewhat accurately is beyond the computing capabilities of your typical laptop computer, so such models are always highly reductionist, deliberately so.
Source: I work with 2D meshes to models and even a single physical object can get almost infinitely deep and complex if you want to make it so.
A Rhodes has a set of metal tines that get plonked like a ruler on a table edge, rather like the tines on music boxes, and there’s a set of two tines, each with its own pickup for EACH NOTE on the keyboard!
Of course, you can use a sampled Rhodes and it’s only noticeable to either the player or if it’s not buried in a mix of a zillion other tracks, there is a notable lack of depth to it.
Like most things on computers: “good enough” wins the day.
I switched to purely virtual synthesis in 1999, and have spent the last 5 years revisiting physical musical instruments as I lament the electronification of music I was such a fan of for 20-30 years
> Physically modeling a Rhodes somewhat accurately is beyond the computing capabilities of your typical laptop computer, so such models are always highly reductionist, deliberately so.
This really not true anymore.
Pianoteq includes several e-piano models and it runs just fine on a typical laptop. It also includes physically modelled acoustic pianos, harpsichords and several other instruments. All run fine on a typical laptop.
The fact there are differences between a software and hardware synth doesn't mean that:
- you could spot it isn't the real one outside of a comparison test.
- you would prefer the real one in a blind hearing test
- it matters once the sound has past through effect, the composition has been mastered and compressed.
- it matters once lossly streamed, playing on the typical shitty bluetooth speakers and headphones while in a noisy environment which accounts for the majority of the context where records are played.
You're evaluating an instrument purely from the perspective of the listener and not the musician.
Sure, an emulated Rhodes might be indistinguishable once it's gone through the audio production and mastering meat grinder. But the real question is whether a musician will even want to put one through that if the sound doesn't inspire them the moment they sit down and play it.
I use a mixture of analog and digital synths. While I appreciate the power and flexibility of emulations, I've never been inspired to make music as much as I am when I press a key on an analog synth.
Surely a serious musician is going to want to create music that rewards attention, not music expressly built for distracted people on shitty speakers they can barely hear.
> Surely a serious musician is going to want to create music that rewards attention
Not always. Consider Robert Rich's very serious idea of "egoless" music which isn't intended to command attention but to leave space for the listener's own internal attention and creation.
In his sleep concerts listeners even fall in and out of unconsciousness, presumably experiencing some interesting dreams as well.
> music expressly built for distracted people on shitty speakers they can barely hear
An apt description for many things, from wax cylinders to AM radio to MP3 to to music played on smartphones. Which is to say: all popular music since the 1890s, "serious" or not.
I routinely hit 100% of a core on Pianoteq on a 3 GHz i5 from a couple years ago (keep in mind it uses multiple cores). Pianoteq is definitely CPU-intensive. Diva hits around 50% and it's also considered CPU-intensive.
If your assumption is that people are always buying brand new Threadrippers or Mac M1s every couple years, then we're living in different worlds.
Musicians, especially, are much slower to upgrade hardware and software than other users because of the risk that something breaks during the upgrade--there are just so many different pieces of software that can break. A lot of musicians with Macs stayed on Mojave for a few years after it became obsolete just so they could run 32-bit programs, even though the writing has been on the wall for 32-bit programs since the very beginning of the Mac on x86 era.
As someone who spent thousands of dollars over the years on software instruments as well as analog and digital hardware synthesizers, I can very comfortably say that software does not get the full sound that hardware does. Additionally, hardware offers much more creative control in general, a tactile interface customized for the specific way it creates sound. The only downside is the hardware is it takes up so much space.
To get the same benefits of hardware in software, it requires a significantly different (for me, much less creative ) workflow, a very large investment of time, and you still will not quite get there in sound quality. But, at least you can take your whole studio with you on a laptop.
I see where you are coming from, but one of the things that I didn’t understand until recently, when I got an OB-6, is how much these synths are music instruments (vs. something you can use to make a track). They are closer to a guitar or a violin than they are to a any of the VST. I love playing the ob-6 way more than any other digital thing that I have owned, but I don’t feel like sequencing it as much.
From the music producer perspective, your point of view sounds very true to me. Sometimes these synths are harder to get in the mix than a digital one, or a VST, as they take too much space! And using them is less convenient than a plug-in. But playing a VST feels much less exciting than one of these analogs. And I am saying this while owning a Virus, a digitone keys and Ableton Live.
Virus is all you need ;-) I have a Virus 2 and Nord Lead 4 as my primary instruments, and their combination is just perfect. Though I think all a regular person needs to explode with joy is just a single Novation Circuit...
Anyone whith a passion for the Viruses and the Nords Leads may be interested in this project that is reverse engineering and emulating the Motorola DSP563xx
https://dsp56300.wordpress.com/
Architecturally, OB-6 and Prophet 6 are very similar as they share a lot of the same circuitry. They are both great synths, and I am sure I'd love a prophet 6 if I had one.
But I always perk up when I hear the Oberheim sound, more so than the prophet sound, and I do like very much the state-variable filter, which is not found in the P6. I enjoy making pad sounds with a notch filter, and be able to modulate it's shape towards a LP and a HP filters.
Compared to the Virus, the OB-6 sounds more alive when played with a keyboard. It's nuanced when played like this, sometimes like an old piano, where the Virus sounds equally full but very precise, which is great when sequenced and you want a bassline that sounds pristine, for example. I find the Digitone a bit in-between, as you can get pretty creative with the FM algos, and it also sounds very hi-fi.
> I don't think there are any advantages today of using an analogue synth.
It's funny. Dave Smith worked on the very first software instruments in the early 90's. He thought it wasn't really fun to use so he went back to creating analogue synths, his first product with his new company was the Evolver, the rest is history.
Turns out a lot of people don't find plugins very fun to use or inspiring, it's as simple as that. Smith was right since some people are willing to pay almost $5000 for a new Prophet 10 Rev4 in retail.
>I don't think there are any advantages today of using an analogue synth.
There is a huge one - unique sounds!
You will see that comment over and over from pro's who make their living creating sounds for TV/movies - this guy's channel is fascinating if you really want a peek "under the covers" https://www.youtube.com/c/NeilParfittMusic
Found him for the Mac Pro review when it was rare and new, stayed for the fascinating insight into the pro music world and the mind of an ultra audio geek!
If you're serious about making music with keyboards, there are a lot of reasons why you would be spending anywhere from $800 to $3,000 on gear just for your basic setup. At these price points, you have the option to get analogue gear, and the tradeoff is that the analogue gear is less flexible (it's not more expensive).
I'm not trying to pick on people for using inferior gear, but explain that the people who spend more money on a keyboard are making very rational decisions. Just to pick your Pianoteq example, once you pair a $150 Pianoteq license with a $500 decent quality weighted MIDI keyboard, you're already in a price range where you could get good analogue gear if that's what you want.
Once you're spending $1k, $2k, $3k on gear, your main questions are going to be about things like ergonomics, personal taste, capabilities, etc. Not everyone is okay with using a laptop, getting a good laptop for use on stage has its own problems, and laptops add to the price. This isn't really that different from the amount of money you'd spend to get into other instruments, and is actually kind of cheap when you compare it to something like a concert violin--you'll even see high-school students with $5k violins, and I'm not talking about kids with rich parents, either.
Again, I'm not saying that there's something wrong with going for a $100 keyboard and your $180 license for u-he Diva. That's an excellent combo. Just trying to explain that there are reasons someone would want to spend much more money in the first place (ergonomics, portability, keyboard feel, etc.) and at those price points, analogue is an option.
I have plenty of software synths. I just bought a Hydrasynth Deluxe (which now sits below my Novation Summit). It's 100% digital, and could easily be implemented as a plugin. Yet it is a vastly better and more rewarding experience for sound design and performance than any soft synth can ever be.
As yet, no one has managed to create a controller for soft synths that is sufficient to give them this versatility. (IMO: They would need an OLED display on every control, and a traditional keyboard with poly-aftertouch, and a template and control standard to reconfigure it for common soft-synths. NKS doesn't even come close.) I'm sure it will happen, but so far it has not.
