Hello members,

I always used hardwaresynths, but I may "downgrade" to software-only because I like the flexibility much. However, I tried many synths, heared many demos but I couldn't find a softsynth that produces really thick and fat unisono-leads (I'm creating trance).
Because I'm also a programmer, I created a little app that creates unisono saw leads with up to 64 voices of any length. The program is batch-operable, so I generated a wave file per MIDI note and created some patches for a software-sampler, so I can use it like a synth with many oscillators.
The results are quite good I think , you may listen to a little demo that I've made:
http://www.nilsschneider.de/temp/kvrdemo.mp3

The demo is using 10 second wave files per MIDI-note, each wave has been created with 12 saw waves. The oscillators are modeled after my microQ hardware synth ( it took quite some time with a function plotter to find the right equation ), and it sounds quite similar.
The file contains two commercial trance riffs played with my "soft synth" only.

What do you think?

Regards,

Nils Schneider

Have you not tried Superwave? All the presets sound like this.

Unisono is very much a performance thing (in any aspect).

Your examples actually merely use (simple static osc detune) oscillator unisono.

But normally a synthesizer, which plays true unisono, also consideres all the other things of a voice (filters, modulations and so on...).

Additionally there are normally some realtime parameters assigned to real unisono too. For instance unisono detune and unisono spread....

This isn't simulate-able with sampling, because the extreme static of sampled waveforms limits the sonic possibilities of real unisono enormously.

An example: The Virus b/C hardware (which can produce real unisono - free definable count of voices) needs under certain circumstances all the available performance for merely a single note when played heaviely unisono...

But this finally sounds much more *living* and organic, than the examples you showed.

But may be usefull for certain things...

.

Kriminal wrote:

Have you not tried Superwave? All the presets sound like this.

As far as I see, Superwave also don't use real unisono, but merely a kinda oscillator stacking. This isn't real unisono.

.

Nils Schneider wrote:

Hello members, I always used hardwaresynths, but I may "downgrade" to software-only because I like the flexibility much. However, I tried many synths, heared many demos but I couldn't find a softsynth that produces really thick and fat unisono-leads (I'm creating trance). Because I'm also a programmer, I created a little app that creates unisono saw leads with up to 64 voices of any length. The program is batch-operable, so I generated a wave file per MIDI note and created some patches for a software-sampler, so I can use it like a synth with many oscillators. The results are quite good I think , you may listen to a little demo that I've made: http://www.nilsschneider.de/temp/kvrdemo.mp3 The demo is using 10 second wave files per MIDI-note, each wave has been created with 12 saw waves. The oscillators are modeled after my microQ hardware synth ( it took quite some time with a function plotter to find the right equation ), and it sounds quite similar. The file contains two commercial trance riffs played with my "soft synth" only. What do you think? Regards, Nils Schneider

hey nils, they sound nice.. but i wouldn't compare them with a good hw synth. i know you modeled it after the Q and the q sounds a bit more digital and less warm then a virus or a jp. but when i hear this sounds i don't think about if i should sell my virus. they are better then some software-synths but there are some good software-synths too. did you already try the polysix or v-station?

so long
cYrus

here is a sample with 2 synths..

www.solarproject.ch/linked/saws.mp3

it has a bit too much low end.. but anyway you should be able to judge it.

Subjectively, Nils, I think they sound great. I'm interested in further information about this ... is this a commercial product? Please let us know!

Thanks,

(stu.macQ)

Thanks for all your replies!

@Kriminal:

What I don't like at the superwave is that you can hear differences in the filtering between different notes, I guess internally the superwave uses a multisample to generate the saws of the JP8000, and you can clearly hear where the next sample is used at your keyboard if the filter is completely opened.

@jackie&hyde:
I know that those things are problems. Because of that I used 10 second wave files even for the second example and set the starting position to random positions, which gives quite a realistical representation.
Okay, filters are usually per voice, not per note. But I tried some filters and the results are quite good.

@cyrus:

>did you already try the polysix or v-station?
I tried the v-station and I think it sound's awful, and additionally it requires a lot of CPU. I don't know the polysix though (may be worth a try). I listened to your two examples and think they're not bad. Altough the second one suffers from aliasing when the filter is completely open. The attack of the first synth is a bit hard (you hear the clicking), but that's tweakable. Which synths are they? First polysix second nova?

@ stu.macQ:
>Subjectively, Nils, I think they sound great. I'm interested in further information about this ... is this a commercial product? Please let us know!
I don't know what I'll do with it. At the moment it's only a little tool for my private use. I may share it, I'm not sure yet. I'll try to model other oscillators next weekend and see how close I can go to Virus, Nova and Oberheim Matrix 1000 (the hardware synths I have available here).
if there is anyone that has got other waveforms, just mail me.
Maybe it'll become shareware.
A software synthesizer is not possible at the moment, because it uses a lot of CPU, I'm creating the waves with 128x oversampling for maximum anti-aliasing. A low-quality version may be possible though, but I didn't look at the VST sdk yet.

you should be able to do that with Sytrus. You have full control over the unison, you can decide if it should layer oscillators or full voices, you can control the detuning, the phase start, the stereo image, and you can assign virtually any parameter to each of the unison voices.

Hi Nils
I listened to your songs
I loved the song "coming home", with el guitars it is just one of the best songs i have ever heard (Pink floyd going trance)

I suggest everyone to have a listen to it

Keep up the good work.

I didn't listen to your demos [can't where I am] but for my money JX Synth has the best unison since my old KORG DSS-1 or DW8000. It is seriously huge. I'm also working on a new synth that uses EVM's unison prefab which I think sounds heaps better than the one I have been using. I just need to iron out a few major bugs.

jxsynth unison/detune is beyond awful.not even a slight chance to make those sort of sounds.
I for one can't wait for this.....(if it's gonna be a freebie)
ps:make it yellow like the microQ!

and to kriminal.....superwave presets sound like that mp3? u r tone death dude.

-

_starcraft_ wrote:

and to kriminal.....superwave presets sound like that mp3? u r tone death dude. -

No, i was being slightly sarcastic, as in the both sound awfull

Kriminal wrote:

_starcraft_ wrote: and to kriminal.....superwave presets sound like that mp3? u r tone death dude. - No, i was being slightly sarcastic, as in the both sound awfull

sarcasm? tsk tsk...

cYrus wrote:

Nils Schneider wrote: Hello members, I always used hardwaresynths, but I may "downgrade" to software-only because I like the flexibility much. However, I tried many synths, heared many demos but I couldn't find a softsynth that produces really thick and fat unisono-leads (I'm creating trance). Because I'm also a programmer, I created a little app that creates unisono saw leads with up to 64 voices of any length. The program is batch-operable, so I generated a wave file per MIDI note and created some patches for a software-sampler, so I can use it like a synth with many oscillators. The results are quite good I think , you may listen to a little demo that I've made: http://www.nilsschneider.de/temp/kvrdemo.mp3 The demo is using 10 second wave files per MIDI-note, each wave has been created with 12 saw waves. The oscillators are modeled after my microQ hardware synth ( it took quite some time with a function plotter to find the right equation ), and it sounds quite similar. The file contains two commercial trance riffs played with my "soft synth" only. What do you think? Regards, Nils Schneider hey nils, they sound nice.. but i wouldn't compare them with a good hw synth. i know you modeled it after the Q and the q sounds a bit more digital and less warm then a virus or a jp. but when i hear this sounds i don't think about if i should sell my virus. they are better then some software-synths but there are some good software-synths too. did you already try the polysix or v-station? so long cYrus

JP....Virus...warm??

I've put the program one step further: It could even become a realtime VSTi.

You have to know: My current job is to program pixel- and vertex shaders of modern graphics cards, and yesterday I tried to let my Geforce FX 5900 calculate the oscillators in a pixel shader

I didn't believe a graphics card would be that fast for sound processing, but the results are great. I'll download the VST sdk now.....

I swear, I've never abused a graphics card that much, but if it works? Who cares

Nils

yeah from jenuary GPU instruments will be pretty much the way to go.
http://www.bionicfx.com/

upgrading graphic card is 10x cheaper than changin cpu and motherboard.

_starcraft_ wrote:

yeah from jenuary GPU instruments will be pretty much the way to go. http://www.bionicfx.com/ upgrading graphic card is 10x cheaper than changin cpu and motherboard.

10x? I doubt it! I can triple my Athlon850's performance with an Athlon2600 motherboard/cpu upgrade that's $107 including shipping:
http://www.pricewatch.com/h/prc.aspx?i=306&a=5063&f=1

Looks like a 128MB NVidia chip video card would be $41.

consider current GPUs are 3 times faster than current CPUs.....so spending 100$ for a cpu will only get u 1/3 of the performance of a 41$ nvidia card.
if this is true there's no way i will upgarde my cpu this comin year.

They are even faster, if 1/3 only that would mean that you could calculate Pixel Shaders in software with 20 fps if your graphics card does 60.

hey nils, that would be awesome if you could support GPUs

btw. i would like to hear a "supersaw-pad" i think that's the hardest thing for a VST. it really sorts the good VSTs out of all.. all those cheap plugins sounds too repetive when playing a supersaw-pad.

so long
cYrus

Quote:

I've put the program one step further: It could even become a realtime VSTi.

mmh my AMD 1700 would run a 3000-oscillator synth. Is 3000 saws not enough for a supersaw?

>my AMD 1700 would run a 3000-oscillator synth

Based on what calculation? And what sound quality?

Quote:

Based on what calculation?

one could argue that a supersaw could be more than a stack of saws, but this would be based on the use of a saw wavetable & some linear interpolation.

gol wrote:

Quote: I've put the program one step further: It could even become a realtime VSTi. mmh my AMD 1700 would run a 3000-oscillator synth. Is 3000 saws not enough for a supersaw?

with many of todays vstis u r lucky if u can get past 32 unison voices.(32 x 3osc =96) with an amd1700 i guess the only synth where u can managae more than 32 voices unison would be pro53.
but 3000? lol

a supersaw is generally 3 saws osc with 16 voice unison...and detuned to the max.

Quote:

with many of todays vstis u r lucky if u can get past 32 unison voices.(32 x 3osc =96) with an amd1700 i guess the only synth where u can managae more than 32 voices unison would be pro53. but 3000? lol

That was my point. He seems to claim that a supersaw is made of a huge lot of voices and needs a GPU to be performed(??), while it's just from 4 to less than a dozen oscillators.

The reason synths don't offer more than a dozen voices for unison/supersaw isn't a CPU limit, it's just that it sounds great with less than that.

Quote:

amd1700 i guess the only synth where u can managae more than 32 voices unison would be pro53.

CPU-wise, you can stack A LOT of oscillators with a cheap AMD 1700 you know. Just check using an additive synth how many of them you can stack (I did using the beepmap).
Feature-wise, I can stack 9*2*6 = 108 saws per voice in sytrus, and it's ok on my CPU (if there's any point in doing that, that is).

gol wrote:

Quote: with many of todays vstis u r lucky if u can get past 32 unison voices.(32 x 3osc =96) with an amd1700 i guess the only synth where u can managae more than 32 voices unison would be pro53. but 3000? lol That was my point. He seems to claim that a supersaw is made of a huge lot of voices and needs a GPU to be performed(??), while it's just from 4 to less than a dozen oscillators. The reason synths don't offer more than a dozen voices for unison/supersaw isn't a CPU limit, it's just that it sounds great with less than that. Quote: amd1700 i guess the only synth where u can managae more than 32 voices unison would be pro53. CPU-wise, you can stack A LOT of oscillators with a cheap AMD 1700 you know. Just check using an additive synth how many of them you can stack (I did using the beepmap). Feature-wise, I can stack 9*2*6 = 108 saws per voice in sytrus, and it's ok on my CPU (if there's any point in doing that, that is).

Well said.

I can tell you from my experiences, that an (real) unisono voice, which means, that you consider the entire voice and not only the oscillators, sounds good between 2 and 8 voices stacked together.

Only very rare *real live* situations may require a higher amount of useful unisono.

And please consider the entire voice with it's filters and all kind of applied modulation. Also the possibility to drive the amount of stereo spread and unisono detune and full stacked automation and possibly other features in realtime ...

An "endless stacked sawtooth wave" (even inside a video card) is not equal nor compareble with the unisono mode of an high-end synthesizer for obvious reasons.

And this is meaned soundwise.

.

gol wrote:

The reason synths don't offer more than a dozen voices for unison/supersaw isn't a CPU limit, it's just that it sounds great with less than that.

my cpu can hardly handle 32 unison voices on some synths (try the minimoog demo for example )....and its a athlon2400! so ur 1700 would hardly handle 16.....try it out!(on the minimoog demo remember to switch on all 3 osc).
or u can try zerovector demo.....with 6 voices unison....50% cpu! (it sounds amazing though).
maybe u hate the idea of GPU audio cos u probably have to recode all your software....but as a consumer i honestly can't wait!


@Nils ....u allready capable of creating a GPU vsti ? when do u think we can try it? or u have to wait for bionicfx technolgy ?

cheers.

Quote:

Only very rare *real live* situations may require a higher amount of useful unisono.

I only use 9-osc saws for really smooth 'shhhh' pads

Quote:

And please consider the entire voice with it's filters and all kind of applied modulation

Depends, I think that a single voice's filter applied to a stack of oscillators doesn't sound much different than the stack of voices themselves. Of course you then miss the stereo spread if they're mono. Now a distortion on the stack sounds A LOT different.

Quote:

my cpu can hardly handle 32 unison voices on some synths (try the minimoog demo for example )....and its a athlon2400! so ur 1700 would hardly handle 16.....try it out!(on the minimoog demo remember to switch on all 3 osc).

It only means that the plugin is poorly optimized

Quote:

maybe u hate the idea of GPU audio cos u probably have to recode all your software....but as a consumer i honestly can't wait!

You can't wait to have a somewhat unpredictable latency added? Or different qualities depending on your graphic hardware?

I personally think the GPU would/will have a good use for impulse reverbs that are way too heavy for our CPU's, but I can't think of anything else. Maybe an additive synth, but surely not a subsynth.