KVR Audio

aMUSEd · Post by **aMUSEd** » Sat Sep 10, 2011 1:01 am

Hunter wrote:As far as I can see all this shows is that nobody has read the Alchemy manual and figured out how to get the best results from it's resynthesis, which offers a lot more potential than the other two? There are various import options, as well as options in pure additive import, there are the options using the spectral engine to complement/instead of the additive engine, depending on the type of sound you are importing:

http://www.camelaudio.com/alchemymanual ... /#Additive

and if you use them properly, I'll be surprised if you can beat it. After that, there is a whole lot more you can do in Alchemy once you have imported a sample than you can do with the other two?

Was going to say the same. Alchemy has more options, more control and 4 point morphing

But I do also love Poseidon for it's ease of use and good results

ENV1 · Post by **ENV1** » Sat Sep 10, 2011 1:16 am

himalaya wrote:we can't compare to the original

This is a precise quote of what you have said.

Where i come from 'i cant' equals 'i dont have the means'.

Now you suddenly claim you had the file all along.

So excuse me for being slightly confused...

himalaya wrote:Don't you hear Vertigo's lowest note all mushy, and Harmor's soft attack when compared to the Alchemy version? Your 4 way speakers should be able to play it back?

For the definitely last time: I did not bother to look for such things!

I have listened to the comparison and posted what i thought sounded best, based on my spontaneous impression, without analyzing [expletive] anything!

Now that thats (hopefully) clear, could we please drop the subject now?

There is nothing to gain from keeping this on and on and on.

Ben H · Post by **Ben H** » Sat Sep 10, 2011 1:21 am

george wrote:This is a simple audio re-synthesis engine comparison between discoDSP Vertigo, Image Line Harmor and Camel Audio Alchemy [snip] Alchemy notes had to be lowered two octaves to get the same pitch.

Did you tell Alchemy what the correct root note was?

Because this can have a huge bearing on how close to the original it would sound.

bmanic · Post by **bmanic** » Sat Sep 10, 2011 1:24 am

himalaya wrote:Ahh, that explains it. Great info, bManic. So, the same stereo result could also be accomplished in Alchemy by loading the Left side into source A, and Right into source B, both imported using the Spectral engine. I wonder how the stereo image would sound when recombined back in Alchemy? I can't try it now as I have to get some sleep. But yeah, I did get a very good resynthesised mono version of your original (cool track BTW, is that one of yours?).

I will definitely check out Harmor, when time permits, it looks and sounds additivitastic (except for that mashed attack on that bass sample!).

Well I just "wasted" 2 hours in Alchemy again.. damn that synth is like crack for me. Once I start tweaking even a tiny bit I get lost in it. What a fabulous synth!

Yeah, that snippet is from one of my millions of unfinished songs. I think that's about as far as I've come with that track.

Cheers!
bManic

bmanic · Post by **bmanic** » Sat Sep 10, 2011 1:29 am

Hunter wrote: If you want to import a sound and hear it exactly like it was originally, perhaps you should be looking at a sampler instead rather than a resynthesis engine

The reason why it is IMPERATIVE to get as close to identical results as possible is so that you can START your morphing from something that resembles the original audio. Of course you can "fake" this in a DAW environment with crossfades and stuff but using it live is a bit more tricky (dj mixer with crossfade can come in handy).

Once we get resynthesis to fully capture an original source of audio, only the sky is the limit. So in my opinion the quality of the actual reproduction is one of the most important aspects, if not THE most important.

If you just want to morph and mangle stuff beyond recognition there are millions of other ways to do that as well.

Cheers!
bManic

Image-Line · Post by **Image-Line** » Sat Sep 10, 2011 3:43 am

rj0 wrote: "Performance vs Quality:

... <Other settings> ...

Precision: Computation precision. Select from:

- Average ... so don't have much impact on your CPU load.

- High - ...

- Perfect - ... require vast amounts of CPU (mainly to be resurrected via simulation) ... Use of this setting should only be attempted if your have access to CPUs not yet invented."

------

Changing from Average to High might be good.

Perfect, on the other hand ...

You missed the reasons...

Precision - Computation precision. Select from:

Average - where all computations are made by a team of college-level mathematicians. They like to play 'Hunt the wumpus' in their spare time so don't have much impact on your CPU load.

High - All calculations are completed by university level mathematicians. This team have special needs and consume large amounts of pizza, cola and CPU for various special interest projects.

Perfect - All calculations are done by Isaac Newton, Pierre de Fermat, Carl Gauss & Albert Einstein. This team of 'prima donna' mathematicians require vast amounts of CPU (mainly to be resurrected via simulation) and partially because Einstein and Newton are endlessly arguing about space-time. Newton just can't accept that it's useful to travel faster than the speed of a horse so Einstein's mathematics are silly and not relevant. Use of this setting should only be attempted if you have access to CPUs not yet invented.

There's a lot going on back there...

tony tony chopper · Post by **tony tony chopper** » Sat Sep 10, 2011 5:15 am

4 pages & you all missed the obvious..

Harmor has formant preservation. If you started with the "resynthesis" preset, formant preservation is enabled. Formant preservation isn't a defect, it's a feature. That's why the other notes of the bass sound more filtered.
You can switch it off if you wanna compare to those 2 (unless they too feature it).

Also, resynthesis isn't about 1 sound, it's tricks for specific cases. Try resynthesizing

-vocals. There you will understand why you need formant preservation.

-drumloops. There you will be able to test transient preservation.
(the same way, if you're not using material full of transients, turn this off)

-mixed material would be better if you only wanna make 1 test. Because if you only wanna mangle with bass or monophonic material, granular synthesis will do as good.

Another tip: if you wanna compare X, Y & Z, don't use samples that came with Z, use samples that aren't in either X, Y or Z. What are the chances that a sample that was included in Z, doesn't sound good in Z? (these are fuzzy algos, test it like you would test a timestretcher, you can always find material that sounds better in one or the other)

(& as others pointed out, there's a higher-quality import setting, which generally doesn't matter, but will make a difference for bass. For a bass you'd typically switch that on, set formant preservation off, and lower the image sharpening)

adenozin · Post by **adenozin** » Sat Sep 10, 2011 5:22 am

aMUSEd wrote:Alchemy has more options, more control

Sorry but the spectral synthesis is quite limited in term of control.

tony tony chopper · Post by **tony tony chopper** » Sat Sep 10, 2011 5:36 am

First you drag the sample to one part, then to the other and it automatically selects the other channel. Thus you get a stereo file.

more precisely (after loading the resynthesis template preset, but you already did that), part A will import the mid and B will import the side (thus the part A/B mix will be a mid/side mix, you can add effects to just the mid or the side, etc). Thus if you only load part A you get mono, and with part B you add stereo. It's all in the "side" switch in ADV, it's switched on in the resynthesis template.

aMUSEd · Post by **aMUSEd** » Sat Sep 10, 2011 7:14 am

adenozin wrote:
aMUSEd wrote:Alchemy has more options, more control
Sorry but the spectral synthesis is quite limited in term of control.

It still has more resynthesis options than the others though (to clarify I meant in terms of import here, not what you can do afterwards) in that you can do additive, spectral or a mixture of both and as I said has the ability to morph between 4 sounds as well

tony tony chopper · Post by **tony tony chopper** » Sat Sep 10, 2011 7:23 am

pretty much everything in Harmor is a "resynthesis option", except the effects in the effects tab. See, I don't known, the resynthesis->effects preset, & compare it to the original sample.

I mean that's the whole point of Harmor, it tries to do as much as possible in the freq domain, all the tools are mangling partials directly.

Krakatau · Post by **Krakatau** » Sat Sep 10, 2011 7:39 am

Hunter wrote: If you want to import a sound and hear it exactly like it was originally, perhaps you should be looking at a sampler instead rather than a resynthesis engine.

I have to disagree,

i for one, think that a, so to say "accurate resynthesis" would imply a better preservation of the sonic characters of a sound source after some treatments that aren't frankly intended to affect them, like dynamic time-stretch controlled by velocity to give you an example

It could make a significant difference, i'm really persuaded...

aMUSEd · Post by **aMUSEd** » Sat Sep 10, 2011 7:46 am

Krakatau wrote:
Hunter wrote: If you want to import a sound and hear it exactly like it was originally, perhaps you should be looking at a sampler instead rather than a resynthesis engine.
I have to disagree,

i for one, think that a, so to say "accurate resynthesis" would imply a better preservation of the sonic characters of a sound source after some treatments that aren't frankly intended to affect them, like dynamic time-stretch controlled by velocity to give you an example

It could make a significant difference, i'm really persuaded...

Yeah I don't agree with that part - I really want to be able to morph between sounds that start as something that is recognisable even if what they become is something new or transitional between two or more recognisable sounds. That for me is the main point of morphing when it comes to resynthesis (ie like Kyma)

HunterKiller · Post by **HunterKiller** » Sat Sep 10, 2011 8:30 am

Vertigo definitely sounds more analog than the other two.

himalaya · Post by **himalaya** » Sat Sep 10, 2011 9:13 am

ENV1 wrote:
himalaya wrote:we can't compare to the original
This is a precise quote of what you have said.

Where i come from 'i cant' equals 'i dont have the means'.

Now you suddenly claim you had the file all along.

So excuse me for being slightly confused...

You confuse yourself:
"we can't compare" meant that we, as in all of us here (not "i") in the thread could not compare the files properly as the original wasn't available with the resynthesised files. And since I have Alchemy, I could've compared then but didn't have the time or the will. Then, after a few posts I managed to get a minute spare and loaded the sample to confirm that the attacks were indeed better in the sample and that Vertigo and Harmor made a mess of it.

ENV1 wrote: I have listened to the comparison and posted what i thought sounded best, based on my spontaneous impression, without analyzing [expletive] anything!

But you did overanalize it even by opening the files in a spectrum analyzer?

Here:

ENV1 wrote:
This is without doing any material analysis, only going by ear.
All i did was checking the spectrum ...

Vertigo vs Harmor vs Alchemy