PlanetZ forums - scopeusers.com

It's that morph synth vsti with the triangle.
http://www.eisenberg-audio.de/einklang

It's a sad purchase really, wasn't sure why I even bought it. Their demos were not very conclusive, so I had to hear for myself to get a better understanding of their tech. After messing with it for an hour, I think it's fairly clear to me what is going on. I think it's a spectral synth that uses fft resynthesis or something similar. And all interesting things that make a synth useful have been ripped out. There are no LFO, no EG, no mod matrix, nothing.

You have some minimal controls, but they're not really what I'd call sound creation tools. Just Attack, Release, Percussion (which is like decay and sustain combined), dissonance, modulation (pitch mod), etc. Breaking ADSR really makes is useless. Because "Percussion" ties decay to sustain, you can't turn sustain to 0 and have a decay of, say, 3000ms for a piano sound or whatever. That is really, really messed up. The single LFO it has is hard wired to pitch for vibrato. Apart from the broken ADSR and LFO, this thing has no temporal modulators, which makes its sound very, very static. It feels like I'm messing with a 2mb gm soundfont.

The morphing part works as expected. Anyone with experience with additive resynthesis and spectral (fft, stft) resynthesis won't find any surprises. I suspect it's a single analysis of a single sample spread across the entire key range. There were some velocity response, but nothing really popped out as being surprisingly velocity responsive. (as in timbre change) It's like a simple single sample rompler but using spectral resynthesis instead.

The sound packs were somewhat interesting. They're organized into colors, sold separately or all together as a bundle. The content of the packs are completely hidden before purchase, other than some general categorical info like "leads", or "ambient pads". Basically, the samples are either of real physical objects/instruments (cello, wurlitzer, etc), or of something synthesized (saw wave with a sweeping lpf with some resonance). Combining real sounds with synth sounds yielded sounds with a bit more of a bite and higher harmonic content. Otherwise the acoustic sounds are all very muffled and unimpressive. Severely lacking in body and heft, not sure what happened there. Morphing synth sounds with each other yielded much more interesting results, but largely depended on the sample content. Some sounds had filter sweeps with LFO which were interesting, others were just plain static sounds. I had to remember which sounds had interesting LFO movement and combine it with static sounds to make it do interesting things.

That's where the whole "you do the music, we do the math" part breaks down. Since there are no parameters to just make your sound, and because all sound is made from a combination of sounds, the whole sound making process is about memorizing what preset (root sound) contained what sort of characteristic.

Let's make a comparison. Let's say a typical synth, that from the creator's perspective is a massive, complicated, scary thing that destroys all creativity, has 50 knobs and faders. Mentally, you map sound characteristics to these knobs and faders. To increase/decrease treble content you control the cutoff on the LPF. That's a lot, but it's 50 things to remember. Now compare it to the morph approach. There are a total of about 480 "root" sounds, each with distinct characteristics. To get a particular sound, you need to break it down to 1, 2, or 3 "root" sounds, each time picking from the mother set of 480 sounds. (480, 479, 478) 480 things is a lot to remember. Not impossible, but a lot. And the morph result is difficult to predict. Maybe it's a learning curve thing, so at some point you can pick 2 sounds, and imagine in your head how a 50% morph between the two may sound like.. might require more practice. But anyway, with 480 things and combinations that create less than predictable sounds, it requires a lot of browsing and messing around in hopes of randomly coming by a cool sound. To me, that's much, much more time spent fiddling with the synth and less time writing.

sounds like they tried to do what wolfgang palm did with plex2 that uses the same technique but has better control

and its free

http://plex.hermannseib.com/

btw, in Alchemy you can spectral morph or additive morph between 4 targets, and have all mod capabilities.

Most of us have enough virtual synthesizers now to last the rest of our lives. Another one isn't going to make your music any better. I would love to see more development in samplers. Acoustic sounds are richer if you have a good microphone and preamp.

btw, I actually tried the quad additive morph setup on Alchemy and it works pretty well. It gets the bubbly artifacts typical from additive analysis/resyn combo, but largely depending on source material. The morph works well, mostly if you do it very gradually with LFO. Doing it too fast makes it sound no different from fading 4 sources. I tried the spectral morph as well, but that had too much artifacting.

I wasn't really ever a "evolving pads" kinda guy, so in general my writing doesn't take advantage of them.. I'm thinking I'll explore it a bit. For now, the results I'm getting are not too different from building evolving sounds from a typical VA setup. Probably need to find the correct type of source material to morph.

Slow morphing brings out the effect the most, but everything around it (in arrangement) also needs to bring attention to it for it to be obvious, otherwise I think it's easy for the effect to get lost in the mix since it's quite subtle. It's an aspect I've never really looked into so it's fresh and interesting.

The einklang thing was a total bust, but at least it was a segway into all this morphing pads related stuff.

morphing i think sanning wavetable synthesis is the best

and why i love my asr10 as you can layer 8 of them for massive morphing sounds

but you can get the same results from vector synthesis

the roland s760 was grest for this as you could do vector synthesis with 4 partial layers

same set up as the jd800 but with better samples as the internal ram was 4 times the size of the jd800 rom

wuzik station is a softsynth sampler that can do it also

and the mach5 3 can do it to

I agree. Whether it's additive resyn or fft, seems like the results are not that much better than wavetable or vector.

we use additive resynthesis for the wavetables it creates single cycle spectra over time

then it uses interpolation from a resynthesis of 33 singles over a time period then fills the gaps with additive spectra

so the new stuff really is only doing the same as the original ppg

the only difference is that it renders per note in a multisample type sense

where as the ppg used the same wavetable for all keys

the good thing about alchemy is that

additive resynthesis cant do noise or transients

so they created algo that gets the noise content of the sound file and any transients from the sound file then layers all 3 together so that you get a better aproximation

but you can do the same just by adding pcm samples for noise and transients

the k5000 has this ability you can have upto 6 layers so you can use pcm transients and then use the other 5 additive osc and envelopes for bring in and out over time to simulate the sounds

and why the roland d50 was a hit it used transients and virtual analog synth section to trick people into thinking it was a brass instrument

in fact the d50 was the first produced va synth the second was the k2000 with its vast engine

jhulk wrote: ...in fact the d50 was the first produced va synth the second was the k2000 with its vast engine

V.A.S.T. engine: Variable Architecture Synthesis Technology

yep its dsp algo

and i still own mine since 1990

and still love the sound 64mb ram when most were on only 32 max for another few years

i still make krz pcm files

single cycle samples and the dsp you can get some great patches as you can layer 3 of them

it was way ahead of its time as was the k250