http://news.harmony-central.com/Newp/20 ... on-VS.html
More fire for the software debate...?
neuron SOFTSYNTH coming...!
imho a sign of financial trouble - though that would be the last thing I'd wish for them...
more important is the aspect that the modelling part, the one which actually contains the neural network IS NOT included.
imho the original Neuron has one of the best 'hands-on-interfaces' to sound - something you can hardly reproduce in software.
I've mentioned it numerous times (and will do again
) that a very similiar synth architecture exists on SFP: the PythonPro - dunno if it's still marketed by Zarg.
check specs here
For a 'perfect' Neuron fake you'd need 2 instances with the output routed into a Modular patch to process (ADSR, filter etc)the result of each 'Resynthator'
cheers, Tom
more important is the aspect that the modelling part, the one which actually contains the neural network IS NOT included.
imho the original Neuron has one of the best 'hands-on-interfaces' to sound - something you can hardly reproduce in software.
I've mentioned it numerous times (and will do again
) that a very similiar synth architecture exists on SFP: the PythonPro - dunno if it's still marketed by Zarg.check specs here
For a 'perfect' Neuron fake you'd need 2 instances with the output routed into a Modular patch to process (ADSR, filter etc)the result of each 'Resynthator'
cheers, Tom
-
Joxer the Mighty
- Posts: 175
- Joined: Mon Nov 03, 2003 4:00 pm
yes, I do - it's one of SFP's hidden gems 
unfortunately it's a bit difficult to understand 'intuitively', unlike the familiar oscillator, asdr, filter scheme.
it's really good in faking semi-acoustic stuff, like epiano, guitar, brass etc - not exactly physical modelling a la Karplus/Strong, but an approach with similiar results regarding the audio part.
the oscillating part is done by 4 OP FM, which is a very effective way to generate a wide variety of sound characters.
then there's a resonance and feedback section, like the corpus of a physical instrument.
The routing capabilities are somewhat complex, as even the final fx stage can be fed back into the resonance parts.
afaik John Bowen did a lot of the interface work and imho he fully succeeded, even though the thing IS difficult to grasp.
But that's in the nature of the concept - you don't have endless screen space.
The Neuron operates with a similiar scheme, but got it's superior hardware surface giving you a more direct access. Even this interface has been described from 'surprising' to 'extremely difficult' by reviewers
You just slightly touch a joystick and the sound jumps right off from a decent pad to a full crash...
cheers, Tom - questions welcome
unfortunately it's a bit difficult to understand 'intuitively', unlike the familiar oscillator, asdr, filter scheme.
it's really good in faking semi-acoustic stuff, like epiano, guitar, brass etc - not exactly physical modelling a la Karplus/Strong, but an approach with similiar results regarding the audio part.
the oscillating part is done by 4 OP FM, which is a very effective way to generate a wide variety of sound characters.
then there's a resonance and feedback section, like the corpus of a physical instrument.
The routing capabilities are somewhat complex, as even the final fx stage can be fed back into the resonance parts.
afaik John Bowen did a lot of the interface work and imho he fully succeeded, even though the thing IS difficult to grasp.
But that's in the nature of the concept - you don't have endless screen space.
The Neuron operates with a similiar scheme, but got it's superior hardware surface giving you a more direct access. Even this interface has been described from 'surprising' to 'extremely difficult' by reviewers
You just slightly touch a joystick and the sound jumps right off from a decent pad to a full crash...
cheers, Tom - questions welcome
Hi Tom, everyone -
Thanks for the nice 'promo' regarding the Python Pro, but I didn't do the user interface graphics at all. Everything was done by the synth's designer, Paul van der Valk. He also did the presets, owner's manual, and the Website listed above for the Python Pro specs. All I did was help assemble the "insides" for Paul, since he didn't have Scope DP to use.
It took me a lot longer than I expected, and it is a very different type of synth - wonderfully complex, and capable of some unusual sounds. I don't know if Paul would characterize it as a Neuron-style synth, since it does not generate anything from samples, nor does it analyze anything from spectra, etc. Basically you have 2 FM 'oscillators', and 2 tube/comb filters, with LOTS of ways to interconnect them, etc.
Please check Paul's Web site, and be sure to send him any questions you might have regarding his design.
And yes, it is still for sale on my site! (check the Zarg Music Store page at the bottom).
Cheers,
_________________
john bowen
zarg music
<font size=-1>[ This Message was edited by: johnbowen on 2004-08-14 05:30 ]</font>
Thanks for the nice 'promo' regarding the Python Pro, but I didn't do the user interface graphics at all. Everything was done by the synth's designer, Paul van der Valk. He also did the presets, owner's manual, and the Website listed above for the Python Pro specs. All I did was help assemble the "insides" for Paul, since he didn't have Scope DP to use.
It took me a lot longer than I expected, and it is a very different type of synth - wonderfully complex, and capable of some unusual sounds. I don't know if Paul would characterize it as a Neuron-style synth, since it does not generate anything from samples, nor does it analyze anything from spectra, etc. Basically you have 2 FM 'oscillators', and 2 tube/comb filters, with LOTS of ways to interconnect them, etc.
Please check Paul's Web site, and be sure to send him any questions you might have regarding his design.
And yes, it is still for sale on my site! (check the Zarg Music Store page at the bottom).
Cheers,
_________________
john bowen
zarg music
<font size=-1>[ This Message was edited by: johnbowen on 2004-08-14 05:30 ]</font>
thanks for the background info, John
So if someone considers the Neuron highly interesting, he should at least have a close peek at the Python - the final result of the synthesis may not be that different
The Neuron's sound generation (and as such the software plugin) is in no way related to neural network technology. So there's nothing spectacular (or innovative) in this domain.
What IS innovative is the audio analysis and encoding of the software that accompagies the hardware Neuron. This is an extremely sophistcated approach which could (simplified) be described as something in the convolution domain, which finally results in a 'traditional' waveform played back by an engine similiar to a Vectron or a Wavestation.
So the Neuron 'wireframes' the sound parameters from sample analysis (by the above mentioned software), while the Python uses a set of FM operators for that purpose.
Everyone knows that FM may not yield the most realistic image of a certain instrument, but it's very good in working out the basic character of a sound.
Since this basic source is heavily modified and spiced with the more or less unusual way of adding resonance corpus characteristics (as does the Neuron), there's not much in commom with the Yamaha DX-ish way of FM - though I admit I'm a fan of the latter, too
cheers, Tom
<font size=-1>[ This Message was edited by: astroman on 2004-08-14 15:43 ]</font>
neither would I - but both synths happen to have a few strategies in common, even if the 'sources' for sound are indeed very different.On 2004-08-14 05:24, johnbowen wrote:
...I don't know if Paul would characterize it as a Neuron-style synth, since it does not generate anything from samples, nor does it analyze anything from spectra, etc...
So if someone considers the Neuron highly interesting, he should at least have a close peek at the Python - the final result of the synthesis may not be that different
The Neuron's sound generation (and as such the software plugin) is in no way related to neural network technology. So there's nothing spectacular (or innovative) in this domain.
What IS innovative is the audio analysis and encoding of the software that accompagies the hardware Neuron. This is an extremely sophistcated approach which could (simplified) be described as something in the convolution domain, which finally results in a 'traditional' waveform played back by an engine similiar to a Vectron or a Wavestation.
So the Neuron 'wireframes' the sound parameters from sample analysis (by the above mentioned software), while the Python uses a set of FM operators for that purpose.
Everyone knows that FM may not yield the most realistic image of a certain instrument, but it's very good in working out the basic character of a sound.
Since this basic source is heavily modified and spiced with the more or less unusual way of adding resonance corpus characteristics (as does the Neuron), there's not much in commom with the Yamaha DX-ish way of FM - though I admit I'm a fan of the latter, too
cheers, Tom
<font size=-1>[ This Message was edited by: astroman on 2004-08-14 15:43 ]</font>