Classic Synth Cores?

[frieza]

Anyone ever look into doing something like this? Recreating synths like the Prophet-5, VCS3, CS-80, etc? Could be interesting having a few FPGA dev boards with 25 or 49 key USB midi keyboards hooked into them to create a similar experience to native hardware without having to spend the literal $1000s per unit that you normally would.

[FatSlob71]

Doing all those chips would be a nightmare! lol good thought though i sold a Roland Jupiter 8 for 10 000 now they are 75 000 lol

[LamerDeluxe]

The thing is that those are analog synths and FPGA chips are digital, so you won't get the same result. Digital synths would be possible to recreate, but possibly not better than a VST recreation.
FPGA chips are very interesting for synths though, as they can create waveforms at such a high frequency that aliasing isn't a problem anymore.
EDIT: That said, the SID emulation on MiSTer is really good and I'd love a SID synth with something like eight of them, with MIDI/OSC support.

[softtest9]

I know that Jotego once toyed with the idea of an MT-32 FPGA core. I don't know if this is something that he will explore further in the future or not.

I feel like I should mention that MAME has a number of synth drivers. Most of them are very underdeveloped but some of them seem pretty good. Not FPGA but it might be more attractive for some people than messing around with VSTs or old hardware.

[virtuali]

Analog synths, which will likely be the most sought after, aren't probably suited to FPGA. However, there are still some classic digital synth that might be interesting, like:

Yamaha FM series, like the DX-7 or the TX81Z.
Roland D-series which are almost all ( except for the D-50 ) based on MT-32. A proper MT-32 in FPGA that could possibly be connected to the Mister, might convince some users to invest in a 2nd DE-10

Casio synths might be interesting too, especially the iconic VL-1, the first commercial digital synth ever.

[Bas]

N00b question: the C64 SID is an analog synth, is it not? How is that different in implementation from something like a classic Moog?

[LamerDeluxe]

N00b question: the C64 SID is an analog synth, is it not? How is that different in implementation from something like a classic Moog?

SID is a hybrid synth. The oscillators are digital, the analog filters are digitally controlled. Audio quality is not as good as that of professional synths. Emulation gets really close, but as there is a lot of variation between SID chips. On MiSTer you can now select multiple filter curves to simulate this.

Fully analog synths don't have any aliasing. Earlier analog synths use voltage controlled oscillators, filters and envelopes. This creates all kinds of imperfections, which makes the sound richer when more voices are playing at the same time. Later models are digitally controlled, which makes the sound less organic. Artificially adding these imperfections per voice to digital emulation does help a lot. Analog filters also add imperfections, which make the sound more interesting. Also high analog levels won't just clip but distort in their own way.

[Moondandy]

Presumably as a first step someone would want to do something very basic, and not very exciting to people e.g. a Casio VL Tone, or maybe an early digital drum machine like the Roland 505 or 626. Even then I don't know if elements in these early digital synths which haven't been dumped or are hidden on chips.

Hoping for something advanced as a first synth seems like expecting the first MiSTer game core to have been the PS1 or Saturn, before we even had a pong core. Would be very cool to have synth cores though, that could be huge for the project in terms of interest.

[thorr]

Agreed. Meanwhile I am using the Arturia V Collection. It's fantastic. I mainly wanted it for the Yamaha DX7, but there are tons of other great synths as well. Don't buy it at $599 though. It goes on sale every so often and usually during black friday. I also have a Roland Fantom 8 that has a built-in Jupiter 8, SH-101, Juno 106, and JX-8P. If any or all of this could be done in the MiSTer with a USB midi cable attached to it, that would be very cool.

[syltefar]

I dream about open source implementations of classic digital synthesizers and samplers. We already have near perfect emulation of a bunch of synth chips in the consoles and arcade machines, including the type of Yamaha FM chip that is used in the DX-7, so it doesn't seem completely far fetched. The MAME team have worked on a few of them, but most don't work yet https://wiki.mamedev.org/index.php/Synthesizers .

[FatSlob71]

JP-8080 would be next to impossible to do!!!

[FatSlob71]

Microwave XT and Virus Ti core please!

[cathrynmataga]

Doing all those chips would be a nightmare! lol good thought though i sold a Roland Jupiter 8 for 10 000 now they are 75 000 lol

Owe, problem is that it's probably best to own one to properly emulate it.

What would emulation of an analog synth even look like? Analog parts aren't really modular, not at all. So if someone designed a transistor component, a capacitor, an inductor component all according to specs, and hooked them together, would be interesting to see if it was possible to make them behave like the real thing. This is like EE grad school project level, something like this.

[Newsdee]

Maybe this could be converted into a core, and instead of using physical knobs have a screen-based interface:

http://www.ucapps.de/midibox_sid.html

You would probably still need a MIDI keyboard for it though (unless you simulate it via a keyboard).

But there is also the question of why not use a regular computer core with software synth, if MIDI is supported for it?

For example:
https://www.youtube.com/watch?v=mp_KNaRAHHM

[cathrynmataga]

People are thinking about FPGA emulation of analog/mixed system. Random google found this. These guys are saying FPGA is 2-3 orders of magnitude (100x 1000x faster) than CPU based, but that it's hard to do. The base components would be pretty challenging to create and have them be accurate, but if we had a library of analog parts and a way to hook them together, then we could build a variety of vintage analog devices, radios, organs, synthesizers, old TVs, and have them actually work. The effort to do this would be pretty vast.

https://www.researchgate.net/publicatio ... it_Designs

"In this paper, we showed that emulating an AMS chip design on an FPGA can lead to a speedup of 2-3 orders of magnitude as compared to existing CPU-based simulations. Unfortunately, creating synthesizable AMS models is difficult and requires a combination of expertise that is rare to find: knowledge of both analog design and FPGA design"

It says open-source in the title, but I don't see an obvious link to the code for what they have so far. Hmmph.

[pgimeno]

It says open-source in the title, but I don't see an obvious link to the code for what they have so far. Hmmph.

Try here:

https://github.com/sgherbst/anasymod

[dshadoff]

The 1980’s would be the obvious place to start, where digitization existed in a primitive form, but created a ‘personality’ for the music.
Somebody’s working on at least one quintessential chip here: (although controls are a separate discussion).
https://twitter.com/kenshirriff/status/ ... 41409?s=20

[CartoonDonkey]

Ok....GO:

https://www.righto.com/2021/11/reverse- ... x7_28.html

[afx]

There are already a couple of new FPGA-based analog modelling synths in the Waldorf Kyra and UDO Super 6. Looking at the prices though, I'm not sure why you wouldn't go for a real analog synth. They're certainly comparable in price to some of Dave Smith/Sequential's desktop and keyboard analogs.

[FatSlob71]

UDO Super 6 and Novation Peak/Summit there would be Legal issues with those but older synths such as the Nord Lead/Virus Ti Microwave XT could be doable but no one is working on any! Jotego could do one ?

[jopdorp]

This is ineresting, I've been working on some mixed digital/analog modeling in arcade cores.
It's a lot of work, but possible.

I'd consider implementing one classic synth.

I'm thinking it would be great if we find some generic way to implement analog audio circuits, maybe similar to SPICE circuit simulation, or maybe some parts can be accelerated by using high-level-emlation/simulation of subcircuits, to save processing power/logic resources.
Another idea would be to do an all-modeled simlation with modular subcircuits that you can bind together.

Anyway. It's not entirely clear to me how the generic solution would work, but maybe we can assamble some kind of analog sound task force that will be able to do this together

[jopdorp]

People are thinking about FPGA emulation of analog/mixed system. Random google found this. These guys are saying FPGA is 2-3 orders of magnitude (100x 1000x faster) than CPU based, but that it's hard to do. The base components would be pretty challenging to create and have them be accurate, but if we had a library of analog parts and a way to hook them together, then we could build a variety of vintage analog devices, radios, organs, synthesizers, old TVs, and have them actually work. The effort to do this would be pretty vast.

https://www.researchgate.net/publicatio ... it_Designs

"In this paper, we showed that emulating an AMS chip design on an FPGA can lead to a speedup of 2-3 orders of magnitude as compared to existing CPU-based simulations. Unfortunately, creating synthesizable AMS models is difficult and requires a combination of expertise that is rare to find: knowledge of both analog design and FPGA design"

It says open-source in the title, but I don't see an obvious link to the code for what they have so far. Hmmph.

I've started doing this in the MiSTer Discrete project
https://github.com/jopdorp/MiSTer-Discrete

Also I've started a Patreon to be able to push all of this to the next level!
https://www.patreon.com/jopdorp

I'm going to have a good look at tha anasysmod project too, thanks.
https://github.com/sgherbst/anasymod

[amstan]

If you guys are into modeling synths, especially sound creation. You might want to look at faust. It's a DSP language specifically designed for audio synthesis. It can compile to a bunch of targets, one of them verilog. Here's a nice demo: https://www.youtube.com/watch?v=xQMCDbAZJKs

[jopdorp]

If you guys are into modeling synths, especially sound creation. You might want to look at faust. It's a DSP language specifically designed for audio synthesis. It can compile to a bunch of targets, one of them verilog. Here's a nice demo: https://www.youtube.com/watch?v=xQMCDbAZJKs

Interesting stuff, thanks!

[pgimeno]

It just occurred to me that the Fairlight CMI is not much of an analogue synth, is it? I believe it's basically a computer with ADC/DAC.

[Syntax Error]

Anyone ever look into doing something like this? Recreating synths like the Prophet-5, VCS3, CS-80, etc? Could be interesting having a few FPGA dev boards with 25 or 49 key USB midi keyboards hooked into them to create a similar experience to native hardware without having to spend the literal $1000s per unit that you normally would.

aah, see, i knew i couldnt have been the first one to think : "a Synther Fpga" the size of a mister with 2 to 4 midi ports i can just plug into my STE running cubase or the like, wouldnt that be great

as for what cores would you like ?

"all those synths i couldnt afford in the 80s , 90s and 00s" (for starters) ?

even if some of them were VST-type emulators i wouldnt really cre if its a soapbox sized plug with midiports that slots into an ST an acts like any synth you want ...

[tonywatto]

I'd be happy with a casio vl-tone core for starters