Wednesday, June 15, 2011

Review of Alchemy II - the mysteries of the attack transient

While I may get some keyboard players upset, I have always found being both a guitar player and a keyboard player very complementary although I always have to smile when the latest distorted electric guitar preset comes out and everyone thinks their Jimmy Hendrix on a keyboard. Truth is that what I have always loved about a guitar is how subtle and beautiful a guitar sounds when played by a talented artist. I would not classify myself as one but I can tell you that years of playing have given me a sonic pallet that can't be duplicated by cleverly placed samples. I can place my thumb behind a pick a certain way and get harmonics but I can't really tell you how. You have to feel it and hear it. I did not read a guitar for dummies book. I had to teach my brain by doing it over and over again until it became natural.

So what does all this have to do with Alchemy? Simple. Often, that quintessential aspect of how a musician plays, their style, is much more dictated by that short universe of the attack transient. Much (not all) of the magic seem to lie in that fraction of a second. Now enter the additive world. Additive synthesis is what we call linear. Put simply, it's made of parts that can be added together. Now for the interesting part. A guitar transient is what is called non linear which means that all those partials do a complicated little dance. In fact, I believe that grains are far more applicable than partial in the world of the transient. If the transient is a bunch of partials dancing, the dance does not easily reveal it's secrets. Partials dance in and out like the particles of sub atomic physics. In short, the additive model does not work to well.

The solution of the Akai K5000, a little known but powerful additive synth, was to use both samples and additive models. The transient was created with samples. Alchemy works much the same way and can mix or cross fade a sampled transient. This adds realism.

But here is another problem. One of the benefits of additive synthesis is that one can morph between one sound and another and explore that wonderful sonic universe in between. So far so good right. Well, sort of but to do this one has to connect the partials and if we are talking about transients it's a bit like trying to do a tango on a dollar coaster. Trust me, I gave done a lot of modeling and believe me, you can get some bizarre artifacts when trying to morph sounds and they don't always sound that good. I have re-synthesized many sounds and have reacted as if eating spoiled food, yuck, phooey. I have often felt as if my sonic pallet was assaulted. The secret is to start off with some good material. I always felt that was lacking in Cube. In fact, I always felt it was a collection of samples from experiments rather than an effective library. Alchemy has what seems to me a solid library and of course, one can always use ones own samples but be forewarned, some models just don't mix.

Try morphing a bell into a guitar and you will see what I am talking about. Bottom line this will confuse the re-synthesis engine. Also using AM will do the same thing. Re-synthesize a sin wave modulated by a sub audio LFO and then move it into the audible range. The re-sythesis engine will first interpret it as amplitude modulation and then see it as partials. First they are not there and then they are there like Schrodenger's famed cat or the Cheshire cat of Lewis Carol and said Alice. Here, we have the sonic looking glass. A thought experiment like this will show you that the world of additive synthesis gets stranger and stranger when you take it places it may not want to go.

I will have to have a part III soon. I will be working again with Alchemy soon and I will continue to investigate.

Tuesday, June 14, 2011

First Impressions of Alchemy (part one)

Well, I have Alchemy now. For a long time I have not purchased any soft synths really for two reasons. The 1st is that I really don't need anymore and I am actually trying to reduce what I use. The second is that I wanted to at least have a rudimentary set of synths working on my MacBook Pro. I accomplished the later and I thought I might try my hand again at additive synthesis. My first experience was with Cube and let's just say that that experience did not end well.

I have to admit that there is still an inner Geek within me. OK, yes, I have a Bachelors in Math and a masters in statistics and another less geeky one that shall remain stealthy. I like an element of mystery. Here is the geek part.

1. Fourier
2. Helmholtz
3. Gabor

The first Of these tells use that any periodic (really important word) waveform is made up of simple sinwaves in frequencies that are multiples of the 1st (called the fundamental). The multiples are called partials.

The second tells use that these partials pare how we pwecieve sound. The 1st expressed a mathematical theorum that can be proved. The 2nd is simply wrong. That's will take me to the 3rd but that will have to be in part 2.

So the theory is that you can build any sound from sin waves (partials) so that has a kind of geeky appeal like when synths came out that let you draw waveforms. The problem is that no sound has a waveform that never changes unless a computer or electronics made it. As they say, therein lies the rub and a huge one it is. Well, to fix the problem those who wanted to create a form of synthesis had to divide up time into windows. Now let's now think of the windows as what we call the time domain and the partials the frequency domain. Once one enters into time periods less than the wavelength of the waveform. We enter the sonic twilight zone. Think about it and you might see why.

So it appears to me that Alchemy hides this little dilemma from the user and what Harry Gohs of Virsyn once told me was the dark art of additive synthesis, how to connect the dots, or rather, windows.

Confused, don't be if you stick with Alchemy. No one needs to know how an internal combustion engine works to drive a car. Alchemy hides it and that's good. It's nasty business.

So, Ok, this is all Geeky stuff and why would one want to bother making sounds from sin waves? After all you can sample (record) sounds right? Yes, but there are two advantages to additive synthesis:

1. You can do magic with time.
2. You can morph sounds

Morph just means one sound changes into another. Now some of you might remember the Korg Wavestation which did wave sequencing and cross fading (and very well I might add). Isn't cross fading the same thing? Do you want to follow the rabbit down the hole Alice? Turns out that fir most sounds pitch changes and, and here I become like a Nationwide agent and blow you mind, the pitch change in partials is not uniform. Trust me, morphing and cross fading are different. Alchemy does both of these as well as some really interesting variations. It also makes sense to mix samples and additive synthesis. Kawai did this with there additive synthesizer. It makes sense musically so kudos Camel Audio.

I will leave more of this to part Ii of this but there is a strong tendency to get wrapped up in math with additive synthesis and get list and loose the music in the process. Alchemy avoids this and IMHO is a much better synth than Cube.