The long DIVA thread

Official support for: u-he.com
Post Reply New Topic
RELATED
PRODUCTS
Diva

Post

that already exists
http://en.wikipedia.org/wiki//dev/random

and is implemented on pretty much every OS in some form or fashion

Post

Sleepwalker wrote:
olikana wrote:
Urs wrote:Any recommendations on digital synths that got noise right?
that's the problem urs. none that i know of.
Isnt it so that computers cant create truly random numbers? I have previously done some thinking about it: What if the source modulator was an analog signal or something like that, "out of the box". Some electrical signal, or say the real noise from the soundcard. Theres a true random signal. Could it be tapped and used somehow to control a software noise generator? Metaphoricly like a human hand would use a mouse to shake a virtual box of sand in random patterns.
I have a feeling that it's not the generation of noise that's the problem but how noise challenges the way we've implemented digital audio recording and reproduction. Fundamentally, natural noise changes states both in amplitude and frequencies far faster than any of the sampling rates we use. The Nyquist sampling theorem applies to pitched frequencies, not noise. We seem to be generally satisfied with the way noise is handled in recording and reproduction, but the moment you start altering pitch and time as many sound designers do, the weaknesses of the system wrt noise start to show up (try to significantly time stretch the sound of waves at the beach). I have a feeling we're going to need some new thinking to really handle the processing of noise properly and that might also introduce a new re-spelling of how we record, process and play back digital audio.

That said, I don't find noise in most of the better synths intolerable at all. But it's somewhat ironic that if I want to compare them to my analog gear, my analog gear currently has to go through the A/D stage of my converters...

Post

Sleepwalker wrote: Isnt it so that computers cant create truly random numbers?
In a theoretical sense it's important to maintain the distinction between random and pseudorandom because we can't exactly prove the difference :hihi:. It's possible to prove more things in more narrow applications. Practically, it's safe to assume white noise is 'random' enough.

Post

I wonder if the usual computer-white-noise isn't actually too random.

In graphics, I often work with pseudo-random noise, like Perlins implementation or Worleys cell noise, and what makes those look so "natural" is, that it isn't really random at all and has certain characteristics that we find everywhere in nature.

Example:
Image

Bigger version:
http://www.screendream.de/Site/3d-desig ... re.html#/0

And if you take the classical "turbulence", it is Perlin noise repeated and added in "Octaves" - half the size and half the intensity for each repetition, which makes it pseudo-fractal as well as pseudo-random. Very similar to brown noise actually, which is the most natural sounding one.

I experimented with this in music as well: Notes played by random note values are just that, random.
But if you use Perlin noise/turbulence, you suddenly get results much closer to what somebody would play on the piano. One of my Processing-patches playing piano made my girlfriend ask, if it is the new Keith Jarret album ;-)
Not too bad I'd say...

Written music - in an abstract way - has a big resemblance to for instance stock exchange rate printouts or even the line of mountain ridges...
And this is exactly what those pseudo-fractals look like too.

For noise generation, I am sure that there is value to be found in those formulas as well and that it could sound more natural and have more "character" at the same time.

The code would have to change though for the much higher frequencies needed by audio sample rates - in my tests, where I wrote a noise generator for Usine, you could clearly hear the repetitions - and some other characteristics may need to be modelled differently to sound good, but I'm sure it could be done.

Cheers,

Tom
"Out beyond the ideas of wrongdoing and rightdoing, there is a field. I’ll meet you there." - Rumi
Sculptures ScreenDream Mastodon

Post

Gorgeous image... and I agree about organic systems, they always have some order and are not random

Post

Thanx for this, Tom. Much food for thought.
Perlin noise == Keith Jarret? Funny... :hihi:
Amazing pic! :)

Post

hehehe ;-)

I'd not say that Perlin noise == Keith Jarret - that would be quite bold - but it can show certain qualities of a real player. What usually is missing is the inner coherence and even more: a beginning and an end. But the figures played can be very natural.

Cheers,

Tom
"Out beyond the ideas of wrongdoing and rightdoing, there is a field. I’ll meet you there." - Rumi
Sculptures ScreenDream Mastodon

Post

Also not random, but this looks very organic too:
http://www.yworks.com/de/imageviewer.ph ... t=800&fs=1

(BTW yEd is a really nice tool to create diagrams for any purposes, I use it regularly for many different things)
The factor of cpu power and user brain is always constant.

Post

Any release date for the new DIVA update including loads of goodies ??? Don't remember where I heard it, NAMM or Frankfurt...

Post

Firechild wrote:Any release date for the new DIVA update including loads of goodies ??? Don't remember where I heard it, NAMM or Frankfurt...
We're going to release Diva 1.2 soonishly, but that one doesn't have new features (other than e.g. VST3 support).

Diva 1.3 with Digital Osc, Uhbie filter and sequencer is scheduled for release in June - we hope. There will however be a public beta earlier, maybe end of May.

Post

ThomasHelzle wrote:
...
Thats a cool idear. How random is random ? Great picture !!

:clap:

Post

Stick me down for beta and I'll make some sUper speshial tarnce sounds :hihi:


At what number in the list of people making the same request am I? :roll:

Post

tehlord wrote:At what number in the list of people making the same request am I? :roll:
146% of this subforum visitors ? :)
Murderous duck!

Post

We've lost count... which is why we're doing public betas as soon as possible.

We would already be there if we hadn't had so much stress with Messe and all.

Diva 1.3 isn't feature complete yet, but we can tentatively think about a public beta. Satin does need more work before we can let her out.

Post

The new oscillators have given me serious sellers remorse. As soon as that baby is out I'm back on the Diva train to maek sum tarnceh sownds.

Post Reply

Return to “u-he”