Any thoughts on SPICE circuit modelling for future products?
-
- KVRist
- 168 posts since 18 Mar, 2004
SPICE circuit modelling seems to be where things are going for analog recreations.
Plugin developers like DMG, Cytomic, Black Rooster, and Analog Obsession etc. have been getting great results that people have been comparing favorably to Acustica in terms of realism.
Any thoughts on this? I'd love to see the unique MeldaProduction take on SPICE.
Plugin developers like DMG, Cytomic, Black Rooster, and Analog Obsession etc. have been getting great results that people have been comparing favorably to Acustica in terms of realism.
Any thoughts on this? I'd love to see the unique MeldaProduction take on SPICE.
- KVRian
- 707 posts since 29 Dec, 2016 from India
yup i agree spice can be used for analogue circuit stimulation which is great for analogue "warm" sounds
but melda has its own place and specialty which is high level of flexibility which i think any other developer cant provide(Reaktor is another one)
if you have the right knowledge of a plugin you can do anything
eg if you like a certain plugin consider an saturating vst for example
so its your job to know what it does to a input signal(it is best to pass a sin wave through it at different frequency)
using spice will only provide a hard wired shortcut to a particular thing you need
but melda has its own place and specialty which is high level of flexibility which i think any other developer cant provide(Reaktor is another one)
if you have the right knowledge of a plugin you can do anything
eg if you like a certain plugin consider an saturating vst for example
so its your job to know what it does to a input signal(it is best to pass a sin wave through it at different frequency)
using spice will only provide a hard wired shortcut to a particular thing you need
REAPER, Phase Plant , Unfiltered Audio TRIAD and LION, NI classic collection,......... ETC
-
Chandlerhimself Chandlerhimself https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=318799
- KVRAF
- 1822 posts since 19 Dec, 2013 from Japan
Although I don't hate hardware modeling, I feel there are a lot of companies already doing it. Melda is doing a lot of new things that other companies aren't doing. I like the fact that Melda is unabashedly digital. Of course there are some analog modeling style plugins in the melda collection, but even they are digital under the hood. Personally I'd rather leave the analog modeling to other companies and leave the digital stuff to Melda.
My Youtube page https://www.youtube.com/user/GuitarChandler
-
MeldaProduction MeldaProduction https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=176122
- KVRAF
- 14339 posts since 15 Mar, 2008 from Czech republic
Well, I'll be honest here - no plans to do anything like that whatsoever. Personally I think circuit modeling is a nonsense. For several reasons:
1) It implies that the analog hardware is "the best we can do", which is an obvious nonsense. Nothing is the "the best". And the analog gear is actually far from the best. The designs are good, but not because it is analog, but because the devs were highly limited, so they needed to work hard to come up with a well functional yet simple design.
2) It implies that all the analog hardware sound the same, which they don't.
3 and main) It assumes that the only way to get the "good sound" is by simulating the original hw the way it works. Imagine you love old trains with steam engines. Now you buy a new Tesla and say "yeah, it's awesome, but there's something about the steam engines, I'd rather if the car worked like that". So you have many options and certainly the most stupid one is "use the electric power to heat water, which will then rotate the wheels"
.
And here it is the same thing. ANY plugin (and in fact HW as well) is a function of input data (samples) that somehow produces output data. And the theory clearly states that there is a finite algorithm for any kind of finite data, in fact there is infinite number of such algorithms. Trying to simulate the circuits as they were is crazy waste of CPU power, that will never work as the originals (but people like the name "circuit modelling") and it's exactly like electric steam engine. Computers are superpowerful devices, that just work differently and there's no reason to waste them on something like this. At least that is my humble opinion
.
The more modern approach is what we used in MTurboComp for instance - creating an extremely complex digital model (because computers are good with digital, because they are digital
) and then use machine learning to get close to the original analog designs. And machine learning is also what the computers are good at, because they are machines
.
1) It implies that the analog hardware is "the best we can do", which is an obvious nonsense. Nothing is the "the best". And the analog gear is actually far from the best. The designs are good, but not because it is analog, but because the devs were highly limited, so they needed to work hard to come up with a well functional yet simple design.
2) It implies that all the analog hardware sound the same, which they don't.
3 and main) It assumes that the only way to get the "good sound" is by simulating the original hw the way it works. Imagine you love old trains with steam engines. Now you buy a new Tesla and say "yeah, it's awesome, but there's something about the steam engines, I'd rather if the car worked like that". So you have many options and certainly the most stupid one is "use the electric power to heat water, which will then rotate the wheels"
And here it is the same thing. ANY plugin (and in fact HW as well) is a function of input data (samples) that somehow produces output data. And the theory clearly states that there is a finite algorithm for any kind of finite data, in fact there is infinite number of such algorithms. Trying to simulate the circuits as they were is crazy waste of CPU power, that will never work as the originals (but people like the name "circuit modelling") and it's exactly like electric steam engine. Computers are superpowerful devices, that just work differently and there's no reason to waste them on something like this. At least that is my humble opinion
The more modern approach is what we used in MTurboComp for instance - creating an extremely complex digital model (because computers are good with digital, because they are digital
-
- KVRist
- Topic Starter
- 168 posts since 18 Mar, 2004
A lot of good points raised.
I doubt there'll be a shortage of developers new and old going this route, so it makes sense for Melda's focus to remain with an unapologetically modern approach to plugins and doing things impossible with old hardware.
There have already been countless models of vintage gear, most have been lacking, but we do seemed to have entered a new era where smaller developers can more easily compete in with the likes of Acustica and UAD (if not on accuracy, at least with character/vibe that draws many to analog).
There are definitely drawbacks to the approach and it's often harder to deviate from the schematics or avoid the limitations of the originals.
Personally I'm more excited about the approach being taken by Cytomic with The Scream and DMG with the new Track Comp, which allow tweaking actual component values and parameter ranges beyond the originals (I don't know if U-He Re-Pro uses SPICE, but it's a hell of a lot of fun being about the tweak the circuit board!).
DMG is also avoiding copying the look of the originals and going with a modern interface. It would be interesting to compare MTurboComp with the new DMG Track comp as they both model some of the same hardware, but with a different approach (SPICE vs. pure algo).
Once people are done spitting out SPICE-based emulations of everything with a schematic, I'd like to see more creative offerings with a hybrid, or modular approach that makes it easy for users to create their own custom devices without needing to get into the nitty gritty of SPICE themselves.
I doubt there'll be a shortage of developers new and old going this route, so it makes sense for Melda's focus to remain with an unapologetically modern approach to plugins and doing things impossible with old hardware.
There have already been countless models of vintage gear, most have been lacking, but we do seemed to have entered a new era where smaller developers can more easily compete in with the likes of Acustica and UAD (if not on accuracy, at least with character/vibe that draws many to analog).
There are definitely drawbacks to the approach and it's often harder to deviate from the schematics or avoid the limitations of the originals.
Personally I'm more excited about the approach being taken by Cytomic with The Scream and DMG with the new Track Comp, which allow tweaking actual component values and parameter ranges beyond the originals (I don't know if U-He Re-Pro uses SPICE, but it's a hell of a lot of fun being about the tweak the circuit board!).
DMG is also avoiding copying the look of the originals and going with a modern interface. It would be interesting to compare MTurboComp with the new DMG Track comp as they both model some of the same hardware, but with a different approach (SPICE vs. pure algo).
Once people are done spitting out SPICE-based emulations of everything with a schematic, I'd like to see more creative offerings with a hybrid, or modular approach that makes it easy for users to create their own custom devices without needing to get into the nitty gritty of SPICE themselves.
-
MeldaProduction MeldaProduction https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=176122
- KVRAF
- 14339 posts since 15 Mar, 2008 from Czech republic
Hehe yeah, I can imagine tweaking virtual electronic components may be fun, after all I played with electronics a lot when I was younger. But now it just doesn't seem to interest me at all anymore... I like the embrace the digital future
. It's the future anyways. Unless there's an apocalypse 
-
- KVRian
- 509 posts since 28 Oct, 2012 from Argentina
MeldaProduction wrote:Well, I'll be honest here - no plans to do anything like that whatsoever. Personally I think circuit modeling is a nonsense. For several reasons:
1) It implies that the analog hardware is "the best we can do", which is an obvious nonsense. Nothing is the "the best". And the analog gear is actually far from the best. The designs are good, but not because it is analog, but because the devs were highly limited, so they needed to work hard to come up with a well functional yet simple design.
2) It implies that all the analog hardware sound the same, which they don't.
3 and main) It assumes that the only way to get the "good sound" is by simulating the original hw the way it works. Imagine you love old trains with steam engines. Now you buy a new Tesla and say "yeah, it's awesome, but there's something about the steam engines, I'd rather if the car worked like that". So you have many options and certainly the most stupid one is "use the electric power to heat water, which will then rotate the wheels".
And here it is the same thing. ANY plugin (and in fact HW as well) is a function of input data (samples) that somehow produces output data. And the theory clearly states that there is a finite algorithm for any kind of finite data, in fact there is infinite number of such algorithms. Trying to simulate the circuits as they were is crazy waste of CPU power, that will never work as the originals (but people like the name "circuit modelling") and it's exactly like electric steam engine. Computers are superpowerful devices, that just work differently and there's no reason to waste them on something like this. At least that is my humble opinion.
The more modern approach is what we used in MTurboComp for instance - creating an extremely complex digital model (because computers are good with digital, because they are digital) and then use machine learning to get close to the original analog designs. And machine learning is also what the computers are good at, because they are machines
.
- KVRAF
- 10174 posts since 16 Dec, 2002
Just to play devils advocate, MAmp is advertised asMeldaProduction wrote:
1) It implies that the analog hardware is "the best we can do", which is an obvious nonsense. Nothing is the "the best". And the analog gear is actually far from the best. The designs are good, but not because it is analog, but because the devs were highly limited, so they needed to work hard to come up with a well functional yet simple design.
High quality analog simulation
-
- KVRian
- 915 posts since 2 May, 2015
...and "Just to play devil's advocate"...whenever I make adjustments on my Twin I have to do it digitally...
.../s~
mba m2 15" | 16gig.ram | 1tb ssd | macOS 26.1 Tahoe
logic 11.2.2 | reaper 7.75 | cubase 14.0.4
focusrite.2i2 | A&H CQ18t
logic 11.2.2 | reaper 7.75 | cubase 14.0.4
focusrite.2i2 | A&H CQ18t
-
- KVRist
- 132 posts since 23 May, 2017
...MTurboComp:
Seems he's officially talking about one side but using the other side for his marketing. I think noone would complain about anything if there wouldn't be his missing backbone. Unfortunately these are the opposites that keep me away from any MeldaProduction software.The ultimate (not-only) compressor simulating most of the vintage analog compressors.
-
- KVRian
- 509 posts since 28 Oct, 2012 from Argentina
I suspect several of you are missing Vojtech’s point: he is rejecting SPICE circuit modelling, not modelling digitally the desirable characteristics of some analog gear. (But not its circuitry.)
-
- KVRAF
- 10376 posts since 2 Sep, 2003 from Surrey, UK
I'm curious ...
Q1: If several developers use SPICE to model a particular analogue gear will they all sound exactly the same?
Q2: Should they all sound exactly the same?
Q3: Or, if they sound different, why would that be?
Q4: How will the developers make their products distinct?
Q1: If several developers use SPICE to model a particular analogue gear will they all sound exactly the same?
Q2: Should they all sound exactly the same?
Q3: Or, if they sound different, why would that be?
Q4: How will the developers make their products distinct?
-
- KVRist
- 132 posts since 23 May, 2017
Well as a salesman I'd be interested in the image I left while presenting me to some (maybe future) customers. As I remember there recently already was a thread where people explained why they wouldn't buy MeldaProduction software (mostly the UI or the developer behind). I'm pretty sure he's aware of that so criticism won't harm him.steve2KVR wrote:...well that was helpful...(sigh)...
-
- KVRist
- 66 posts since 19 May, 2007
Marketing slogans are always a simplification and all companies do that since marketing simply works that wayDocSnyder wrote:...MTurboComp:Seems he's officially talking about one side but using the other side for his marketing. I think noone would complain about anything if there wouldn't be his missing backbone. Unfortunately these are the opposites that keep me away from any MeldaProduction software.The ultimate (not-only) compressor simulating most of the vintage analog compressors.
Here on the forum, Vojtech explains his perspective on this subject and his attitude is stable over time. I can't see any "missing backbone" - rather the opposite!
I think his approach is not in conflict with his marketing. He just doesn't use the exact analog emulation as we know it from the other companies. Sorry, but your post sounds like you are basically complaining that he reveals what is under the hood...
