Rendered at 17:26:44 GMT+0000 (Coordinated Universal Time) with Cloudflare Workers.
franky47 22 hours ago [-]
I was lucky to be given a lecture on FM synthesis by John Chowning himself during his visit of my home town about 20 years ago. As a kid that grew up playing a Yamaha DX11, that was a great moment.
hananova 12 hours ago [-]
The name FM always bothered me. You're not modulating the frequency, you're modulating the phase. PM would be a better name.
If you were to modulate frequency, well first of all you'd have to define a reference point or you'd just get glitchy sound. In practice that means most FM implementations just have a highly oversampled single cycle of a sine wave in memory, and it's the speed at which the sample pointer consumes the looping sine that's set to correspond to some frequency, and then that value is modulated. In other words, phase modulation.
boomlinde 6 hours ago [-]
Phase modulation is essentially
sin(phase + modulator * modulation index)
For a non-DC modulator without discontinuities it does result in an effective modulation of the frequency. You can think of phase modulation as a means to achieve frequency modulation with certain desirable properties that distinguish it from other methods.
6 hours ago [-]
gugagore 10 hours ago [-]
d/dt phase = frequency
that matches
> it's the speed at which the sample pointer consumes the looping sine that's set to correspond to some frequency
The rest of your message doesn't make sense to me. Both phase and frequency can be modulated, perhaps most clearly seen with FSK vs. PSK.
bradishungry 14 hours ago [-]
I had a few brief conversations with John a few years back when I was helping a friend out at a synthesizer convention. What was extremely interesting to me is that he seemed relatively uninterested in FM synthesis from a musical point of view, at least he made it sound like at the time that was not his interest, and he was just very interested in a computer science point of view. The musicians I was with were very surprised about that. I could kind of understand it as a computer science student at the time and a musician, but it was still really surprising that such a monumental contribution to music was created by someone who wasn’t really focused on that at all!
TheOtherHobbes 8 hours ago [-]
He worked with a synthesizer programmer and patch developer named Dave Bristow while at Yamaha. There was a lot of mutual respect and a certain amount of mutual incomprehension, because Chowning wasn't really thinking about FM musically. Nor were the original Yamaha engineers. The predecessor to the DX series, the GS-1 and GS-2, initially had a lot of harpsichord-like sounds.
The DX7 was a hit because of the polyphony and touch sensitivity, but also because the presets - especially the synth basses and e-pianos - were so obviously musical and immediately useable.
But then you listen to the music Chowning was making, and you can already hear a lot of the tone quality. It's academic electronic music, so no bass and e-piano, but it's pretty impressive for 1977. So it's clear he was musical in his own way, but more not interested in using synths in pop than not interested in music at all.
I can't really understand FM synthesis. Subtractive synthesis is ok, it's just basic signal processing, but for FM I understand the concept but not how you are supposed to predict the sound when it's modulation of modulation of modulation of a carrier wave
someonebaggy 6 hours ago [-]
You don't, you just play with it until it sounds good.
stefanwebb 12 hours ago [-]
This is really interesting! I created a short recently about FM synth with some cool animations that may be of interest to folks: 3 AMAZING facts about FM synthesis
https://youtube.com/shorts/yrZrM2-S5mE?feature=share
ahartmetz 21 hours ago [-]
The first surprise for me is that it wasn't invented in Japan, which I just assumed because all the best-known FM chips and synthesizers are from Japan.
wildzzz 17 hours ago [-]
Many of the early synths came from America. John Chowning actually first licensed his FM synthesis patent to Yamaha so that's probably what gave them a tech boost they needed to grow in the market.
Early electronics from Japan were often seen as inferior products compared to the American and European products (RCA, Westinghouse, Philips, Telefunken, etc). Japan of yesteryear was the China of the 1990s and early 2000s, some decent stuff but a lot of cheap clones. Now, China makes everything so that analogy falls apart today. The 1970s oil crisis was one of the major events that helped bring Japanese goods into a better light, Japanese cars were absolutely killing the domestic automakers on fuel efficiency and price and rather than engage protectionist tariffs (Chinese EVs today, hmm?), they forced the domestic automakers to try harder. This change in public perception helped give a real sales boost to other manufacturing sectors, like consumer electronics which in turn helped drive innovation in a very short time period. That's not to say Japanese product was junk in the 50s and 60s, there were a few dominant players of quality products (Sony) but like Chinese manufacturing today, there was still a lot of cheap junk.
Rochus 17 hours ago [-]
> John Chowning actually first licensed his FM synthesis patent to Yamaha
Did he? Wasn't there New England Digital (NED) with the Synclavier which had FM on the market before Yamaha? Are you sure Chowning gave licenses? I would rather expect that this was the Standford administration's business, not Chowning's.
earthnail 16 hours ago [-]
They offered the patent to all the American synth companies, and they all declined. Multiplying sounds just sounded too ridiculous.
Then came Yamaha, who saw its potential, and made a fortune with it. So did Stanford. That patent became the second most successful patent Stanford ever had until it expired in 1996 (by now obv the university has had way more financially successful patents).
It secured Chowning his tenure; the university probably wouldn’t have prolonged it otherwise. Quite an incredible story the whole thing.
Rochus 16 hours ago [-]
> Quite an incredible story the whole thing.
Even more with Chowning's background as a musician, not an engineer.
> Multiplying sounds just sounded too ridiculous
Is this really the "official" reason, or your interpretation?
ajxs 6 hours ago [-]
> Is this really the "official" reason, or your interpretation?
Stanford did reach out to the American synth manufacturers looking for a company to license the patent. If I recall correctly, the problem wasn't that they didn't get it, the problem was that they weren't interested in, or capable of, building digital synths^1. At that time manufacturing LSI chips for synthesisers was absolutely cutting edge, and Yamaha had taken a major gamble in developing their in-house semiconductor manufacturing capability. Because of that, Yamaha were pretty much the only synth company capable of doing it.
1: The prototypes for the first-generation Yamaha FM synths were built with discrete components, and were enormous. It's understandable that it wouldn't be economical unless you could manufacture your own LSI chips.
Rochus 6 hours ago [-]
> the problem was that they weren't interested in, or capable of, building digital synths^1
Didn't the Synclavier have FM and was on the market even before Yamaha? So I would assume that at least NED must have been interested and have the capabilities of implementing FM.
ajxs 5 hours ago [-]
You're right. NED's Synclavier I (1977) was on the market before the Yamaha GS1, but unfortunately for them, Yamaha had already licensed the patent four years earlier (1973).
Check out how complex the design of the Synclavier is: http://www.studiorepair.com/gallery/misc/NED_Synclavier_II/ (and this is the more modern Synclavier II!) It's huge, and cost a fortune. I'd totally understand a company not wanting to manufacture and market it.
dreamcompiler 13 hours ago [-]
FM is not multiplication. FM is driving one Voltage-Controlled Oscillator (VCO) with another. As Chowning points out it's fast vibrato.
AM (amplitude modulation) is multiplication. When you do it fast, you get what's commonly called a ring modulator.
ahartmetz 4 hours ago [-]
AFAIK, you cannot vary a VCO's control voltage with a frequency similar to its output frequency and get a clean result, so the implementation had to be numeric / digital. But idealized VCOs are a nice mental model for FM synthesis.
JKCalhoun 15 hours ago [-]
Fun to watch the old Kurosawa film, High and Low to see "old", scrappy, Japan.
dreamcompiler 13 hours ago [-]
Agreed.
Can't vouch for this but from what I've read the 2025 remake (Highest 2 Lowest) is...not good.
Lio 11 hours ago [-]
This is not that surprising the big Japanese manufacturers have worked with many American and European engineers for some of their most beloved instruments.
E.g. Akai working with Roger Linn on MPC and Chris Huggett on the S1000 sampler.
Or Dave Smith working with Yamaha and Korg on physical modelling and wave sequencing.
TheOtherHobbes 8 hours ago [-]
Also David Cockerell on the S1000 and Tim Orr on the AX range - both originally from EMS, who gave the world the VCS3 synth.
8bitsrule 14 hours ago [-]
Good to see & hear Chowning talk about the origins!
The DX-7's lack of patch-cord 'programming' ...replaced by 32 instantly-selectable, well-conceived preset patches in RAM ... was essential in 'democratizing' synthesis. Because it wasn't easy to make new patches, it also jump-started the patch-creation business.
The 16-note polyphony also didn't hurt.
" in 1986, it was used in 40% of the number-one singles on the US Billboard Hot 100." -- Wikipedia
chvid 11 hours ago [-]
Haven’t the FM synthesizer patents expired at this point?
You can buy a full FM synthesizer on Temu for very little.
If you were to modulate frequency, well first of all you'd have to define a reference point or you'd just get glitchy sound. In practice that means most FM implementations just have a highly oversampled single cycle of a sine wave in memory, and it's the speed at which the sample pointer consumes the looping sine that's set to correspond to some frequency, and then that value is modulated. In other words, phase modulation.
that matches
> it's the speed at which the sample pointer consumes the looping sine that's set to correspond to some frequency
The rest of your message doesn't make sense to me. Both phase and frequency can be modulated, perhaps most clearly seen with FSK vs. PSK.
The DX7 was a hit because of the polyphony and touch sensitivity, but also because the presets - especially the synth basses and e-pianos - were so obviously musical and immediately useable.
But then you listen to the music Chowning was making, and you can already hear a lot of the tone quality. It's academic electronic music, so no bass and e-piano, but it's pretty impressive for 1977. So it's clear he was musical in his own way, but more not interested in using synths in pop than not interested in music at all.
https://www.youtube.com/watch?v=988jPjs1gao
https://www.youtube.com/watch?v=988jPjs1gao
Early electronics from Japan were often seen as inferior products compared to the American and European products (RCA, Westinghouse, Philips, Telefunken, etc). Japan of yesteryear was the China of the 1990s and early 2000s, some decent stuff but a lot of cheap clones. Now, China makes everything so that analogy falls apart today. The 1970s oil crisis was one of the major events that helped bring Japanese goods into a better light, Japanese cars were absolutely killing the domestic automakers on fuel efficiency and price and rather than engage protectionist tariffs (Chinese EVs today, hmm?), they forced the domestic automakers to try harder. This change in public perception helped give a real sales boost to other manufacturing sectors, like consumer electronics which in turn helped drive innovation in a very short time period. That's not to say Japanese product was junk in the 50s and 60s, there were a few dominant players of quality products (Sony) but like Chinese manufacturing today, there was still a lot of cheap junk.
Did he? Wasn't there New England Digital (NED) with the Synclavier which had FM on the market before Yamaha? Are you sure Chowning gave licenses? I would rather expect that this was the Standford administration's business, not Chowning's.
Then came Yamaha, who saw its potential, and made a fortune with it. So did Stanford. That patent became the second most successful patent Stanford ever had until it expired in 1996 (by now obv the university has had way more financially successful patents).
It secured Chowning his tenure; the university probably wouldn’t have prolonged it otherwise. Quite an incredible story the whole thing.
Even more with Chowning's background as a musician, not an engineer.
> Multiplying sounds just sounded too ridiculous
Is this really the "official" reason, or your interpretation?
Stanford did reach out to the American synth manufacturers looking for a company to license the patent. If I recall correctly, the problem wasn't that they didn't get it, the problem was that they weren't interested in, or capable of, building digital synths^1. At that time manufacturing LSI chips for synthesisers was absolutely cutting edge, and Yamaha had taken a major gamble in developing their in-house semiconductor manufacturing capability. Because of that, Yamaha were pretty much the only synth company capable of doing it.
1: The prototypes for the first-generation Yamaha FM synths were built with discrete components, and were enormous. It's understandable that it wouldn't be economical unless you could manufacture your own LSI chips.
Didn't the Synclavier have FM and was on the market even before Yamaha? So I would assume that at least NED must have been interested and have the capabilities of implementing FM.
Check out how complex the design of the Synclavier is: http://www.studiorepair.com/gallery/misc/NED_Synclavier_II/ (and this is the more modern Synclavier II!) It's huge, and cost a fortune. I'd totally understand a company not wanting to manufacture and market it.
AM (amplitude modulation) is multiplication. When you do it fast, you get what's commonly called a ring modulator.
Can't vouch for this but from what I've read the 2025 remake (Highest 2 Lowest) is...not good.
E.g. Akai working with Roger Linn on MPC and Chris Huggett on the S1000 sampler.
Or Dave Smith working with Yamaha and Korg on physical modelling and wave sequencing.
The DX-7's lack of patch-cord 'programming' ...replaced by 32 instantly-selectable, well-conceived preset patches in RAM ... was essential in 'democratizing' synthesis. Because it wasn't easy to make new patches, it also jump-started the patch-creation business.
The 16-note polyphony also didn't hurt.
" in 1986, it was used in 40% of the number-one singles on the US Billboard Hot 100." -- Wikipedia
You can buy a full FM synthesizer on Temu for very little.