Skip to content

Instantly share code, notes, and snippets.

@antoinevg

antoinevg/synth_history.md

Created August 18, 2022 14:33
Show Gist options
  • Save antoinevg/c752cdb9a283272d1a8bae3509bcc28c to your computer and use it in GitHub Desktop.
Save antoinevg/c752cdb9a283272d1a8bae3509bcc28c to your computer and use it in GitHub Desktop.
Download ZIP
A biased, over-simplified, totally inaccurate and mostly made-up history of the synth industry
Raw
synth_history.md

@bitshiftmask wrote:

If any folks out there are looking into building some kind of audio synth device, something bigger/beefier than a eurorack module, maybe in the league of an OP-1 or a Digitone (or similar), I'd love to see what a 1GHz core and a DSP core could do with some async magic :D

I've kind of given up building synths for lent, but here is my TED talk about the commercial aspect:

There are many synths out there. I mean a lot. Like really a lot. Too many even to keep track of anymore. To understand this it's interesting to look at a biased, over-simplified, totally inaccurate and mostly made-up history:

  1. In the beginning there was the BBC and big piles of re-purposed electronic test equipment. Oscillators, ring modulators, lag generators and the list (USD100 000) goes on. Dr Who featured.

  2. After too many angsty performance art eventually there arose companies like Moog who figured out you didn't need the accuracy and mil-spec designs of test equipment so you could make cheaper versions that talked nicely to each other and so the modular synth (USD 10 000) was born and Keith Emerson's career went into overdrive.

  3. Eventually a standard routing evolved that covered most common use-cases: Ocillator(s) -> Voltage Controlled Filter & Filter Envelope -> Voltage Controlled Amplifier & Amplifier Envelope. Thus was born the Mini-Moog (USD 1 595) and many others like it. Mostly one-voice/monophonic because it was expensive (USD 6 900) to build multiple copies of the circuit for lots of notes and microprocessors hadn't been invented so it was a pain in the ass to decode keyboard signals and round-robin them to each of the voices. Nonetheless it was enough for Wendy Carlos to become a legend.

  4. The first small micro-controllers appeared. Companies like Roland quickly figured out they could control the frequency of many oscillators, generate envelopes and decode keyboard signals at a fraction of the cost of previous polyphonic synth designs. So the Juno (USD 3 000) was born just in time to make the 80's happen and launch a thousand new synth manufacturers.

  5. Then, just a couple years later, things got really crazy to generate the first cambrian era of electronic music technology:

    • Yamaha teamed up with [email protected] to produce the first all-digital synth that consisted of an ASIC containing a sine wave in a lookup table + some envelope generators + 8 copies of a 6 oscillator structure that could multiply sine samples in 32 different topologies or "algorithms" to create the DX7 (USD 2 000) that made possible a) a polyphonic synth that cost a fraction of the price to produce, b) a far far richer range of sounds than the fixed Mini-Moog architecture of the analogue synths.
    • Wolfgang Palm built some digital oscillators that could be controlled from a Motorola 68000 and play back sets of sample data from ROM and interpolate between them in real-time to create the first "wavetable" synths (USD 9 000). I still think they sound horrible but even today they're quite popular amongst a certain breed of German techno producer.
    • Fairlight & others tapped into the (then) unlimited budgets (GB£ 18 000 ~ 60 000) for new music studio technology to just bypass all the fancy and use sheer brute force to sample any sound directly and manipulate and sequence them on their very very very expensive workstations.

  6. Which brings us to the early 90's by which time microprocessors were everywhere, the market was massive enough to make custom ASIC dev cheap and multi-megabyte ROMs no longer broke the bank. Synths simply became big lookup tables hooked up to a DAC. (USD 500 ~ 3000) Basicallly it marked the end of all new synthesis technology for most of the decade. Everything was replaced with boxes full of real instrument samples, session musicians were bankrupted and everyone sold of their massive synth collections. Everything sounded amazing and it was boring as fuck.

  7. Then something weird happened. All those obsolete dirt-cheap analogue synths flooding the market got picked up by young folk abandoned by mainstream culture and a penchant for psychedelic drugs. Turns out even a shitty Roland TB-303 (USD 395) originally meant to accompany your guitar practice can sound massive under a moonlit beach in Goa. By the end of the 90's you could probably buy a small house for even the crappiest piece of old 80's analogue gear. But it got weirder:

  • Christopher Kemper, amongst others, discovered the joy of suddenly affordable DSP dev kits like the Motorola DSP56300 and proceeded to re-implement that old Mini-Moog synth architecture entirely in the digital domain. If you ever want to blame something for Eurotrance blame his Access Virus (USD 2 000).
  • Karl Steinberg found that new chips like the Pentium were getting fast enough that you could do simple signal processing and - at a push - even generate some almost useful bleeps & bloops out of them which led to Cubase and the VST or Virtual Studio Technology™ standard that could basically put a fair chunk of a Fairlight into your homestudio (USD 600 + computer + audio interface).
  • In the quest for even more far out psychedelia a lot of academic research into weird techniques like granular synthesis and Karplus Strong suddenly became commercially viable. This still hasn't really taken off quite as much as it should have.
  1. Then, halfway through the first decade of the next millenium, things started settling down into the status-quo most of the industry has been stuck in ever since:
  • Everything is a VST-plugin running on a PC. (USD 10 - 300)
  • Most VST-plugins are ever more cpu-cycle-consuming emulations of some "classic" piece of gear invented somewhere between the 60's to the 90's which will indue your song with that bespoke "analogue" magic you need to be recognized for the golden-eared genius you are.
  • And everything is a subscription. (USD 5 - 50 / month)
  1. It's not all bad though. The last decade has also seen a lot of cool things coming onto the market such as the Eurorack standard/cottage-industry, dirt-cheap embedded platforms that makes companies like Elektron possible, some truly creative VST plugins and digital filters that don't sound like a wet sock.

So where are we at in 2022?

  • Machine Learning AllTheThings™. For signal-mangling, tubes and (still waiting for the first commercial products) noise-making.
  • Supply-chain catastrophe for anyone needing an ARM processor or audio codec.
  • Ongoing commodization of the PC product space with the most amazing but also worst-named music technology standard ever: CLAP (As in: have you ever had the the ...)
  • Too much studio "hardware" running Linux or some weird version of Windows under the hood that takes ages to boot up and crashes frequently.
  • Needing to either find replacements or pay a massive fee for new versions of your production software every time Apple rolls out a new OS.
  • Sitting through a constant stream of software updates every time you try to switch on your studio computer.
  • Equity funds rushing to buy out everyone with an even vaguely coherent product line, killing half the products and switching to a subscription model for the rest.
  • Multiple competing network audio standards that take days to configure using absolutely abysmal buggy software and require you to switch everything on in a highly specific order to get it working even 50% of the time.

Thing is, there's a pattern here:

Basically the gear industry tends to be generational:

  1. Some technological advance occurs that shifts a design constraint on some noise-making or signal-mangling process. Let's call it "TheNewThing".

  2. Eventually someone in a university, someone at Yamaha/Roland/Korg or (more often these days) someone in a garage builds something with TheNewThing.

  3. More time passes until one day, as if out of nowhere, a producer figures out how to get something musically interesting out of TheNewThing, a chart-topping single ensues and now everyone must have TheNewThing!

  4. Everyone with money puts way too much capital into figuring out how to sell TheNewThing to as many people as possible.

  5. Everyone else figures out how to build TheOldThing cheaper with TheNewThing.

  6. No one fixes the bugs and paper-cuts from the previous generation.

  7. Repeat.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment