Hi there people!
Has anyone ever used Additive Synth by Causal Agency, and gotten any sound out of it? I know I haven't. I dl'ed it ages ago but it didn't work
in cubasis vst3, so I ditched it, and I haven't seen anyone talk about it here...
It looks really interesting though, but is it any good?
anyone?

I tried it out when it was first released as version 0.something. They had a bug that made eat up the cpu cycles - I think they said something about it running in debug mode. When the 'final release version' came out, it didn't work (I think it was because they had spaces in the .dll name) - shame, 'cause it was kinda nice...

Of course, it was a pain to program - you have to set pitch & amplitude envelopes for each partial and you have to know what you're doing. If you do it right, you can get some really interesting warbly bell sounds. It's a really special sound when each partial sine modulates independently. Of course it's gonna be heavy cpu load when you have to calculate 32 accurate, independent sines in realtime, no matter how you code it.

I'm looking forward to trying out the additive synth that DiscoDSP just announced!

What's that Disco DSP synth doing in that screen shot? Looks like it's learning to mimic a wav file sound.

That makes sense, with additive synths being so *%¤#¤%& hard to program from scratch. It would be a great thing to get some initial parameters from a sample and then fiddle with it!

Boy oh boy this vertigo thingie is gonna be exciting!

[quote][/What's that Disco DSP synth doing in that screen shot? Looks like it's learning to mimic a wav file sound[quote]

Simply... yes. It tries to learn the behaviour of most determinant partials along time and apply them to the oscillator bank. At this stage it's even able to 'model' high partial variation waves with not too bad quality, since it uses only 256 partials.

The bad side, "Vertigo" voices using 256 partials, (imagine, is running like 256 oscillators, each one with up to 256 point amplitude envelope), takes a heavy amount of CPU (I'm talking about 28% CPU per voice of a p3 1ghz) and probably we'll release this only for SSE2 processors (p4 or above).

I'll upload some demos soon. (original wav vs modelled)

Personally, I think it's very exciting to see all new products like Neuron, Virsyn "Cube" wich looks awesome, it's the logical next step on synthesis or sound sampling.

This technique is still on early process and it's not perceptibelly good as old and cheap time-domain sampling and of course is very expensive on the CPU process, but imho, the old way it's doomed and it's about a few years we'll forgive all problems concerning time-domain sampling with additive synthesis.

[quote="arguru"][quote][/What's that Disco DSP synth doing in that screen shot? Looks like it's learning to mimic a wav file sound

Quote:

Simply... yes. It tries to learn the behaviour of most determinant partials along time and apply them to the oscillator bank. At this stage it's even able to 'model' high partial variation waves with not too bad quality, since it uses only 256 partials. The bad side, "Vertigo" voices using 256 partials, (imagine, is running like 256 oscillators, each one with up to 256 point amplitude envelope), takes a heavy amount of CPU (I'm talking about 28% CPU per voice of a p3 1ghz) and probably we'll release this only for SSE2 processors (p4 or above). I'll upload some demos soon. (original wav vs modelled) Personally, I think it's very exciting to see all new products like Neuron, Virsyn "Cube" wich looks awesome, it's the logical next step on synthesis or sound sampling. This technique is still on early process and it's not perceptibelly good as old and cheap time-domain sampling and of course is very expensive on the CPU process, but imho, the old way it's doomed and it's about a few years we'll forgive all problems concerning time-domain sampling with additive synthesis.

Arguru,

You guys are doing resynthesis, cool. Don't worry about making it accurate, we've got samplers. Just make it sound great. Oh, and if you get round to it can we have patch morphing assignable to the mod wheel? Sorry to sound like a kid in a sweet shop but I like additive synthesis

Hi Guys,
If you want to hear a cool example of additive synthesis, you should hear the CUBE from VirSyn.
Here is a link to a page that has some CUBE samples.
http://www1.keyboards.de/magazine/m0503/305064wp.html
It's in german, but you'll get the point

So is Additive Synth dead and gone, or what?
Speaking of hard to program: I've got a DX7, so I'm not completely scared off (yet)
Virsyn Cube you say? Hmmm...will it run on a Celeron 800mhz?

Hi Johnny,
Well, here are the system requirements...
System requirements PC:

° Pentium III 600 Mhz processor
° 128 MB RAM
° Windows 98/ME/2000/XP
° 16-bit Soundcard with DirectX/ASIO driver

Hopefully, your system will be able to hack it

Oh, I think I can handle that, just barely!

johnnyluke wrote:

Oh, I think I can handle that, just barely!

Good to hear Johnny.
Be sure to send VirSyn an email if you want to know when the CUBE ships, or just stay tuned to the KvR news...

johnnyluke wrote:

So is Additive Synth dead and gone, or what?

Nope. It should work, it's open source and it had (has) some bugs. There was one causing it to crash on startup on some hosts. I fixed this and sent the changed code back to Causal Agency but I get the feeling that scalar is pretty busy right now so the new build may not be on the site yet. Anyway, I'd have to check the legality of it but I could probably email the build I did to anyone interested? (Works for me in Orion Platinum and Chainer and Massiva demos)

It still doesn't save the envelope stuff correctly but it's worth working with. I understand that in addition to the existing attack envelopes theres going to be a set of release envelopes as well... For the time being it's probably best to setup a sound you like in Chainer and render to a SoundFont...