What chips do synths use?

valis · Post by **valis** » Mon Nov 28, 2005 12:40 am

http://www.sequencer.de/specials/synthesizer_chip.html
^^ Answers

Shroomz~> · Post by **Shroomz~>** » Mon Nov 28, 2005 8:00 am

Thank you very much Valis. We've been trying to find info on the Waldorf Q+ & as soon as I took your link then the Motorolla DSP link, THERE IT WAS in all it's glory !!

Thanks again

darkrezin · Post by **darkrezin** » Mon Nov 28, 2005 3:09 pm

A Neuron is a PC running Linux.

Shroomz~> · Post by **Shroomz~>** » Mon Nov 28, 2005 3:38 pm

Yeh, shame it's not an open source audio developement platform to boot. It could have been so much more when you consider the possibilities of the hardware.

darkrezin · Post by **darkrezin** » Mon Nov 28, 2005 3:51 pm

[off-topic]
Since there's 1 or 2 Neuron users here... can you give me any pointers on this thing? I've tried really really hard to like it (I just about got over the horrible Java app used to create models), but the resynthesis was pretty metallic and thin sounding, and the morphing between resonant bodies sounded like some kind of formant filter effect. I tried to look in the manual but all I got was a load of garbage about spheres and neural networks

I'm willing to be proved wrong and would love to get some use out of it... right now it's sitting in the corner of the studio, with me trying to convince the owner to sell it and buy an analog modular

I didn't pay the money for it so I'm not freaking out or anything - but if I can get any useful sounds out of it, it would definitely be a bonus.
/off-topic

Shroomz~> · Post by **Shroomz~>** » Mon Nov 28, 2005 4:23 pm

Yeh, sorry for the OT valis

[off topic]
We don't own one, a good friend did & he actually loved it. Thought it was the best thing since sliced bread infact

He let us hear some creations from it, but to be honest I thought they sounded average sonically.

But then he sold it ... saying he'd buy another one day ... mmm
[/off topic]

darkrezin · Post by **darkrezin** » Mon Nov 28, 2005 4:32 pm

Shroomz~> · Post by **Shroomz~>** » Mon Nov 28, 2005 4:51 pm

He didn't say as much, but I reckon the plan (& a very sensible one it was) was to sell it for 2 grand or whatever, basically get a good price back for a nearly new one, then buy another in 3 years for <500

darkrezin · Post by **darkrezin** » Mon Nov 28, 2005 4:58 pm

Except they'll probably be collectors items because Hartmann went bust (and they didn't sell a huge amount of them as far as I know).

Shroomz~> · Post by **Shroomz~>** » Mon Nov 28, 2005 5:36 pm

I've posted a 30 second snippet from a Neuron track our friend sent us ... so we can all have a listen to what a synth expert gets from a Neuron. It's in the music forum <a href ="http://www.planetz.com/forums/viewtopic ... 0">Here</a>
Comments on the sound of the Neuron in this example are the point really, because i'm not huge on the style of the track or anything, but then some may love it, who knows

astroman · Post by **astroman** » Tue Nov 29, 2005 3:59 am

well, yesterday I happened to find a new forum opened by an (ex?) Hartmann supporter/sounddesigner to fill the gap in case the Hartmann site gets closed during their insolvency issue.
Some pretty revealing faqs, this one about
<a href=http://www.surroundsfx.com/forum/viewtopic.php?t=10>the Neuron's parameter sets</a> explains what influences the sound quality.
Obviously there was a lot to be improved and faster hardware may have done it, but then the box would have been sellable even worse.

Btw I doubt that 'a synth expert' has any advantages at all to succeed on a machine like the Neuron.

Since a 'sample' is used as the starting point of a physical model, an ambiguos thing to say at least, there's a huge path of abstrahation from source to destination.

Imho the criticism of this system is really undeserved.
It's still the only engine that encodes audio in a different (math)data model to have tweakable sound parameters.

This is not even remotely related to 'neural nets' - though they use it as a buzzword of course (but eventually in vain).

The typical math analyzers of sample content are bound to fail by design concept - a trained ear can do the job at least a 100 times better. Otherwise we'd already have (for example) vocal removers from mixes.

But an open source project ? -lol-
I just aquired a WLAN driver for MacOSX 10.4 - yeah there's tons of open source stuff in this domain and they can do ALL (after someone else did it...)

cheers, tom

darkrezin · Post by **darkrezin** » Tue Nov 29, 2005 5:33 am

Yeah Tom in theory it's interesting, but as I said it just seems a bit gimmicky... resynthesis was not remarkable by any means and the morphing sounds like a formant filter trick. This should not necessarily equate to a bad synth but when I'm using it I really am pretty underwhelmed by it - it sounds like a wavetable synth but without the musical character of a Waldorf.

I have checked out a friend's Kyma system recently, and I can tell you that the sound quality makes your jaw drop... even previewing the resynth analysis processes at 32 partials, the realism is astounding. On the downside, you have to be a bit of a geek to get the most out of it (scripting fun). I'll check out the Neuron demo and support site but I'm still waiting for someone to convince me.

astroman · Post by **astroman** » Tue Nov 29, 2005 1:16 pm

well, the sound quality of the Neuron has never been it's strongest side - but that afforementioned FAQ revealed a surprisingly high demand (or 'number of opportunities') for improvements.
The first sound examples of the Neuron had so much Wavestation 'touch', that I considered the reviewer pretty stupid to not be able to get something else from the machine...

Exactly that seems to be a crucial point: this kind of sound engine IS difficult to handle. The Neuron's counterpart in SFP, the PythonPro hasn't got much attention either, and that is by far no 'bad sounding' synth.

We're so used to think in filters and modulation (some even succeed in 4 FM Ops fluently), but beyond that 'routine' we'd have to start at point zero.
And we're just too impatient to work it out. If an approximation can't be materialized immediately via a preset choice, then we forget about it, don't we ?

So there's a lack of time (or effort) from the users side and the same applies to the developers of the system.
Of course there's never enough time to release such a complex beast in perfect shape, things are underestimated etc.

How long did it take from Modular One to Flexor ?

I'm not into convincing anyone that this is a great machine (it's innovative technological approach misinterpreted as a promise for ultimate sound quality anyway), but it was (?) a great opportunity drowned by ignorance.

cheers, Tom

[ This Message was edited by: astroman on 2005-11-29 13:20 ]

darkrezin · Post by **darkrezin** » Tue Nov 29, 2005 4:31 pm

The support site definitely has some interesting things to try out... I'll have to rig the thing up again to have another go at it.

I think a lot of the reason why I was so disappointed with it was the sound quality.. it didn't 'invite' me to explore it. Something like the Kyma needs *serious* time to be put into learning how to harness that power (8 or 9 analysis techniques for starters - and actually programming it to harness the different parameters and variables is incredibly complex). However hearing that system makes me really wish I had daily access to it to learn it.. I would love to invest an extreme amount of time into this beast.

[ This Message was edited by: darkrezin on 2005-11-29 16:35 ]

Shroomz~> · Post by **Shroomz~>** » Wed Nov 30, 2005 6:32 am

On 2005-11-29 03:59, astroman wrote:
But an open source project ? -lol-

I think you missed my point Tom. The synth in question runs on what? Embedded Linux? Well the sky's the limit in that domain surely?

This beast could have been so much more if it could have been a Neuron one minute & something else the next. They could have built a unit with software on cards as Roland have for their VariOS & V-Synth range.

Anyway I couldn't possibly have a dig at a synth we don't own, so i'm merely pointing out that they may not have squeezed the most from the hardware in that particular design.

In this chapter of pro audio hardware design more interesting concepts are sure to follow.

astroman · Post by **astroman** » Wed Nov 30, 2005 9:36 am

On 2005-11-30 06:32, Shroomz wrote:
...I think you missed my point Tom. ...

maybe not exactly

this hardware controller was designed to support a certain idea, an integral part of the concept.
I see no reason at all why any idiot should be given access to something that's a galaxies' distance beyond his (or her) capabilities.

As a general controller it's way too expensive to open additional business opportunities, imho.

that sarcastic note was of course influenced by that WLAN thing I had to download '...yeah, but there is a free driver, bla...' only to find out the dudes aren't even able to lable screen items in an understandable way.
Of course it did not work at all...
Strangely a 'commerceware version' ($20) did - immediately and without questions

cheers, Tom

ps: can't resist a sidekick now and then in the OS direction - but it's nothing serious
I'm just old fashioned...
wish I had a chip info to drive this to the topic again - shall I open a Casio ?

... ah yes, that's what your friend's example reminded me at: Vince Clark on a Casio quartett

don't underestimate that - I collect them...

[ This Message was edited by: astroman on 2005-11-30 10:56 ]

Shroomz~> · Post by **Shroomz~>** » Wed Nov 30, 2005 12:11 pm

aah yes, chip info.... cheers Tom

Shroomz~> · Post by **Shroomz~>** » Thu Dec 01, 2005 3:43 am

who could dispute the local Neuron loyalty camp

Man if CW & Hartman paid you guys for PR work, you'd be very wealthy indeed & wouldn't have to do any more work for them than you already do

astroman · Post by **astroman** » Thu Dec 01, 2005 5:04 am

On 2005-12-01 03:43, Shroomz wrote:
who could dispute the local Neuron loyalty camp Man if CW & Hartman paid you guys for PR work, you'd be very wealthy indeed & wouldn't have to do any more work for them than you already do

well Shroomz, the economic position of both companies does not exactly express a success of those attempts..., but tnx for your kind words

this approach may have worked around 1990, but today ?

my 'personal' link to artifical neural networks technology also dates back in those years. I've started to get familiar with the technology, but simply lacked the time to do it properly, as all the other developement projects simply left no time.
Some books from this time are still on my shelf - unread.

I was interested in analyzing medical images (ultrasound, cancer detection) in the first place, but a synth project published by a mag also caught my attention.
Intel had just released a 'neural network on chip' that seemed highly promising, and exactly that chip was used as a sound source, producing waveforms similiar to the body's nervous activity.

Yet I've learned enough to realize the vast distance this technology has from the paradigm of 'classical' computer programming.
There's few to nothing in common and that's the main source of my respect for Stefan M. Bernsee, the mastermind behind the Neuron's engine.
It's an enormous mental effort to switch between these 'viewpoints' on the fly, let alone have a product like the Neuron at least 'almost' finished.

cheers, tom

[ This Message was edited by: astroman on 2005-12-01 05:07 ]

Shroomz~> · Post by **Shroomz~>** » Thu Dec 01, 2005 5:18 am

Yeh, sorry Tom. I forgot that neither company could afford to buy you lunch without asking their bank manager never mind pay you