The first electro-mechanical desktop synthesizer | Hacker News Reader

The first electro-mechanical desktop synthesizer

gamechangeraudio.com

47 pointsv-yadli46 comments

I think my electromechanical synth is way more interesting. It's a coat rack with 2 strings with stereo pickup, and played using a rotary magnetic bow, which is an Open Sound Controlled drone motor with a 3D printed wheel with an arrangement of magnets. Only the spinning magnetic field of the bow touches the strings. It is played by subtly controlling the velocity of the wheel. I call it an electroduochord.

Here's a set I played with it. https://youtu.be/nKFK_OhQv3k (Wireless soft-pot controller as well as wireless "hat" controller that uses accelerometers)

And here is the recent design of a rotary magnetic bow wheel with balanced irrational angled arrangement of magnets to reduce harmonic locking. https://www.instagram.com/p/Cr4ZXGztY27/

Here's the OSC Rotary Magnetic Bow code... https://github.com/spDuchamp/OSCRMB

And here is an album made autonomously using an algorithm built in Puredata to control the bow based on the output sound in a feedback loop. https://stefanpowell.bandcamp.com/album/autonomous-drone-lul...

Here is an autonomous recording session... https://www.youtube.com/live/LpQBtJmrez8

I got vibes of Eno's Apollo from the first youtube you shared. Bravo.

Ah, Thanks. An honourable comparison.

If this sort of stuff is your thing, I recommend the work of Trombonist Stuart Dempster, the recordings of Australian composer Alan Lamb, and Trilogie de la Mort by Éliane Radigue.

>HOW IT WORKS

>This is one of the Synth's eight motor oscillators. It can produce four distinct waveforms - the optical disc creates sine, saw and square waves through reading wave reflections via infra-red sensors while the electromagnetic pickups at each motor's base produce the inductive "M" waveform.

The picture shows a small motor driving a physical "strobe" disk printed with 3 different patterns, which are read by optical sensors to create the electrical sine, saw and square waves whose frequency depends on the motor RPM at the time.

The electromagnetic pickups at the bottom of the motor apparently generate an "M" shaped wave as the motor turns. This is the one that's conceptually similar to the original Hammond organ starting from 1935 which also did mechano-electrical waveform synthesis.[0] Although the Hammond motor ran at constant speed synchronized to the power line frequency. The Hammond had a metal disk for each note of the musical scale, precision machined like gears so each note would be at the proper pitch, and the lobes of the disks create the periodic magnetic induction to the pickups which generate the repetitive waveforms.

Definitely not a desktop instrument, a Hammond can be as big as a desk itself, and even heavier. But it was a true synthesizer.

It was also in the 1930's when Hewlett & Packard started up and developed their electrical frequency generator, where a single vacuum tube triode can be wired as an oscillator and used as the source of precision waveforms at a frequency of choice. Making it directly possible for someone to construct a fully electronic organ along the same lines. A musical instrument like this would be a more modern milestone.

So about 20 years later once the patents on these type circuits expired, Thomas organs began to appear in the 1950's where each note had its own triode, so it was a very early organ which was truly a fully electronic synthesizer.

Still didn't sound as good as the Hammonds, but Hammond had decades of absolute top continuous improvement in engineering by then.

[0] Also somewhat analogous to a crankcase position sensor on an internal combustion engine.

> Although the Hammond motor ran at constant speed synchronized to the power line frequency.

However, there must be a way to turn that on and off, to create the characteristic pitch bending effects.

Using motors for synthesis isn't new.

https://en.wikipedia.org/wiki/Tonewheel

Hmm, sounds novel at first. But when I looked at the description of the synthesizer and their videos, I noticed that these are not free-running motors that would do their own thing. Instead, these are brushless DC motors tightly controlled with a feedback loop (that's why they can spin up and down so quickly without noticeable over- and undershoot, which you would hear similar to when you bend a string too much when plucking it).

OK, motors could have spindle play and other imperfections that could make the optical disk tumble a bit and introduce phase effects. Or, as WalterBright wrote, there could be resonances between the motors. Yet, all this tight feedback makes them control the acceleration and deceleration tightly. There is nothing analogue to how these motors are driven. Not an old motor spinning up at its own pace, it's all pre-determined by the motor controller. The motors precisely follows the pre-determined speed curves, from what I hear, similar to how a digital oscillator will. Plus, the envelope is digital anyway.

One of my first thoughts was: Cool, so I can touch the oscillators and skew the pitch! No, I can't. First, there's a cover. ;) But I guess the motor controller controls the pitch so tightly that the effect would not be musically interesting. Compared to when you put your finger on a vinyl deck, for example, which has much larger mass and often a much slower motor control, leading to these slow glides.

Apart from the fancy looks, I really wonder which interesting new soundscapes this instrument can make available. To me, this looks like a "normal" oscillator with ADSR control, just re-labelled as acceleration and brake parameters.

As much as I love the analog sound of early electric guitars and keyboards, I'm okay with simulations. After all, at the end of the day any possible sound is the output of a one dimensional function that takes the time and returns a value. You call this function 40k times a second (at most), wiggle a voltage in proportion to the return value, feed the output stream into a digital-to analog-converter of some sort, and bam! you can generate any sound possible, modulo the capability of your speaker to articulate different frequencies.

The only other boundary condition on the function is that we want to parameterize it in a way that is comprehensible and pleasant for humans to manipulate. I would suggest that a simulation of this synth would be superior to the synth itself in every way. If you were clever enough (or masochistic enough) you could even build in "flakiness" and "parts wear" and "thermal variation".

If you really want an analog sound, then get an orchestral instrument, or an electric guitar! I don't understand this half-measure of "use-machines-to-synthesize-sounds-but-not-THAT-way".

I like them as alternate history retro computing art pieces. They look cool and do in fact make actual music.

I just don't like them enough that I'd ever want to actually own one...

> wiggle a voltage in proportion to the return value, feed the output stream into a digital-to analog-converter

The first part of this sentence, the wiggle, is already a DAC.

You're right, I misspoke. The DAC turns data into a voltage wiggle which moves the magnet which moves the membrane of the speaker. But my point stands, I think.

I feel like someone could generate sounds like this by learning DSP synthesis techniques and dogmatically treating aliasing as a feature instead of a bug.

Edit: to be clear, I like the sounds. I just think DSP beginners learn early to watch out for aliasing in their algos and have the side effect of rejecting timbres that are in any way reminiscent of aliasing from their toolset.

Check out Noise Engineering. Their oscillator modules use a variable speed sample rate so aliasing changes frequency proportionally to the note being played.

So rather than keeping a fixed sample rate with dynamic oscillator pitch (and having to account for bandlimiting the oscillator dynamically as it changes pitch) the oscillator is at a "fixed" frequency but the internal sample rate is changed to get it to play at your desired pitch. This way the aliasing is tuned to the harmonics of the oscillator no matter what pitch it plays.

They also make VST plugin versions of some of their modules but I don't know if they emulate the variable sample rate behavior.

This is impressive and fun but completely impractical, for 2726 reasons. (That's how many Euro they are selling it for.)

A huge market for synthesizers are rich kids who just like blinkenlights. This will sell pretty well.

Yeah...I know a couple of folks with 5-figure $US synth collections who can't actually play. They futz around with the knobs and press a couple of keys and call it 'ambient' or 'trance' or somesuch.

See also: guitar hoarders.

I'd ask that you define what "playing" a synthesizer entails. A lot of that comes down to designing sounds, especially given the fact that you have so many different kinds of oscillators interacting with each other.

Someone can play a keyboard, perhaps, but that's only one of many interfaces to interact with synthesizers.

It's not really the same as a guitar, where your fingers are directly responsible for the vibrations coming out of the strings.

found the rich synth kid.

I've been playing piano my whole life. Recently I bought some synths and it's a whole different world. Piano skill comes in handy sometimes, but other times its' mostly irrelevant. And sound design of course is a completely different skill, which on many synths you can do live as you play.

I've seen people on youtube making great music on a stack of modular gear, without a keyboard in sight, and I can't look down on them because at the rate I'm going it'll be years before I can do what they do.

Yeah...they had to quit fondling their $400 oxygen free silver plated kevlar coated audiophile USB cable[1] long enough to lecture us on the nature of music.

And demand their mom bring them Cheesy Poofs.

[1] No, really...https://www.dedicatedaudio.com/collections/digital/products/...

Maybe one day! I don't own any hardware synths. I do own several guitars, though.

To be fair; I didn't go to a bunch of elite schools so I probably didn't have the same exposure to the level of rich kid that you do/did.

I meant it mostly in jest. Most of the rich synth kids I've met have been rich kids in the rave/burner scene that got so into the music that they wanted to make it but didn't know how to play any instruments, but I'm sure there's plenty of synth hoarders at the fancy schools.

If I had the money I'd probably be one myself. I love blinkenlights and analog electronics and electronic music, but have zero musical talent.

Ehhh, some guitar hoarders just like the different variations. Some actually understand the differences between guitars and have different ones for different sounds. Most studios have at least five different electric guitars not to show off, but because they need them in order to create different soundscapes. The way you can tell who a shiny hoarder is, is that they go for the extremely expensive stuff when a real player would know the cheapest possible option. I've got six guitars, each one has a specific role when recording, and I've never paid more than $200 for a guitar. I've got a friend who has nine guitars, costing between $500 and $2,000, and they can't play anything they didn't learn from a chord chart off Ultimate Guitar.

The same thing applies to synths. Everyone wants a $5,000 Juno-60 when you can get the same functionality out of a cheap $450 CS Reface. Or to go even cheaper, a $100 pre-amp, $100 used PSR keyboard, and access to VST4Free.

There is a type of guitar collector who is actually a hoarder though and doesn't play. Sometimes derogatorily called "Blues Lawyers" after the stereotype of a lawyer who has all the money in the world to buy guitars, but no time to play them because of their profession. Same thing goes in synth communities.

Personally I like these people because they keep my local music shop in business.

Hah, I supposed I'd like 'em too. There's more guitars than people to buy them, and the people who buy for wall hangers usually ended up selling them for a third of their original price after a few years, giving new players access to quality equipment.

I'm obviously not talking about people with multiple instruments because they're looking for a particular effect. I'm talking about people who have 2 Juno-60s they can't really play because it's just money and "I'm a musician...you should see my studio" is a pickup line that worked that one time.

So pretty much the stage 5 of Gear Acquisition Syndrome.

That is not unusual money for an analog synth.

Now here's an earlier electro-mechanical desktop synthesizer. Made from Legos, even.[1]

[1] https://vimeo.com/90101413

I think it'd be cooler if the motors turned slower, slow enough to see, and had more lines on the disk to get the same sound pitch. Optical encoders for motor position sensing often have 1000 lines, so they could spin around 1 rev/second. Add a spiral pattern in the center to make rotation obvious.

I've wanted one of these for years but can't justify the cost. Pretty sure they used to be quite a bit cheaper too and price increased on later production runs. Maybe now the original model will go down in price on the used market.

I can already see modding: print your own "waveform discs".

Also, use the fingers as a break for low frequency modulation.

The idea is brilliant as a novelty synth. Remains to be seen if it's actually useful and expressive enough.

In the YouTube video, the audio lacks definition, like there is a blanket over it.

I suspect it's a cheap trick to fool people who associate bad old recordings with vintage sound.

Demoed by Sonicstate: https://www.youtube.com/watch?v=Q5Y7GQzIatg

So what can a mechanical oscillator do that a digital one can't?

Digital effects pedals have trouble emulating the nonlinear behavior of vacuum tubes, so I imagined the same issue would come up with mechanical implementations. I've never heard a digital trumpet that sounded much like a real trumpet.

Digital implementations also often suffer from being too perfect.

You won't hear a trumpet suitably emulated by motors either. But that wasn't the question.

Any motor-based oscillator can be accurately emulated by a digital oscillator, so the answer to the original question is "nothing." But yeah, neither is good enough to emulate a trumpet played by a human.

The trumpet on my prophesy ain't half bad but yes I concede that point.

For synthetic sounds however, at this point I think it's mostly a question of workflow. Like yes you can take time to program all those imperfections into a digital sound. Or you can take time to make an analog device which has them inherently.

Motors always have imperfections, resonances, feedback from the room its in, etc. I don't know if they'd have enough for it to be audible.

Indeed, motors are physical devices with their own imperfections. But … these are brushless motors, tightly controlled with a feedback loop. They, as they also say in their videos, very quickly follow input changes. I wonder how much weird effects you might hear on those.

Plus the only way to emultate it is to have the real thing as a reference beforehand.

> I've never heard a digital trumpet that sounded much like a real trumpet.

How do you know that every trumpet you've ever heard that sounded real wasn't synthesized?

I used to play a trumpet for years. I had no talent for it, which is why I gave it up. But I know what they sound like.

Screw up! ;)

They literally perform worse than digital oscillators, and that's desirable. Pure sounds are boring, sounds with character are what makes music interesting!

sound like a mechanical oscillator -- up until someone figures out how to emulate it, which is true of every instrument. But, here it is, and it sounds unique, and that's the point.

nothing but they are way cooler than digital oscillators

Reply on news.ycombinator.com