Hi guys... just a quick (I hope... I feel you are very close to haunt me down )
question. Like six months ago I tried to import in additive a distorted power chord. It sounded amazing and I used this method I love to change the sine wave of the partials to others, like saws or like that. And then, modulate the pitch knob (which is in harmonic mode of course).
The question, although, is this: A distorted guitar (or any sound) has a noise component, right? (correct me here, if needed, please)... So I was just experimenting and left it there, but now I just got my guitar, have sonar X1 expanded that comes with guitar rig. Noise components should be analyzed with add+spec don't they? what distortion is everywhere (I can't filter the spec part, but I can't leave it as it is because the spectral part also resynths the harmonic partials, so the "pure" tone of the guitar would be amplified, and the "distortion" sound would be less in volume). Of course I'll check it out myself, but would love some insights on someone who has more experience on this or Camel Guys who knows Alchemy's engine inside out.
I assume that additive engine by itself would be in trouble analyzing the distorted powerchord (which has a fifht in it too).
So, any ideas... I know you can make awesome sounds with this (I did), but I wanted more insights on how this work or should be treaten. I know there is no right or wrong, specially with alchemy, but I hope you know what I'm asking for.

And, would love, if everybody else agrees, on a topic of tricks or methods of usng alchemy, like this I have given, that users and Camel Guys came up in all this time. Not the "how to do this sound" but more like "this technique is like this and does this and it's cool, at least for me..." to share our own experiencies traveling trough this awesome sonic rocket, that has so much room for experimenting. Not an actual sound, but techniques... .
I think would be awesome for this community.

Hugs, a lot!

Mauro... hope you can help me with this, and please think about the last idea, I really think it would be very cool!

Hi,

When analysing a sound using FFT there is a kind of frequency limit above which it becomes very difficult to determine meaningful pitch information.

Spectral analysis is much better at capturing the full range of frequencies from a sample, but it does not provide the same control over the individual partials. This is because the frequency spectrum is divided up in to fixed bands (called bins). Although these high frequencies may be referred to as "noise" they still play an important part in how real we perceive a sound to be.

This basically means there is a trade-off between the flexibility + editing capabilities and the quality of resynthesis but by combining the 2 techniques we can improve the quality of resynthesis and hence our perception of the original sound.

The additive engine captures the main body of the sound and provides control over the partials. This represents the part of the sound that our ears will use to interpret the sound / instrument. The spectral component captures the high frequency "noise", which as explained above is like the icing on the cake when trying to convince our ears that the sound is real.

When the sample has been analysed, Alchemy determines the optimum high pass filter cutoff for the spectral component, which is basically the point at which the pitch information can no longer be meaningfully determined or manipulated. If you were to analyse a sample using Add + Spec mode, and then turn the volume of the additive component to 0 and set the spectral high pass filter to 0 you would hear only the spectral resynthesis.

Does that make sense?

Peace,
Andy.

It does make sense, and sorry to say this (don't malinterpretate it) I already knew this. Maybe my question wasn't clear enough. Will rephrase it: A dist guitar, doesn't have "noise components" like all over it's spectrum? I mean, not only on the highs.
One thing I didn't knew... you say that with the same sample, the additive resynth engine works different if you use it alone and you use it with the sepc too? Like if you take some weight on it, if you know what I mean...
But the spec engine does the same, I mean, it resynths the whole sample (I've always felt this is the best engine in terms of accuracy, except you turn the stretch knob a lot to the left, when it starts to stutter, like a fixed granular engine). So, if I want the dist guitar exactly (and don't want it in sample mode) spectral is the best, but I can't "play" with the harmonic partials (although i could use the noise-resynth mode for a weirder distortion-like effect).

But going to the main part of this, the high pass filter on the spec shouldn't exist, because the distortion is everywhere, not only in the upper frequencies , I mean, of course the fundamental doesn't distort, a distorted fundamental is a sound with that fundamental and another frequencies (harmonic and/or inharmonic).

I wrote this above, here it is (don't know how to quote): (I can't filter the spec part, but I can't leave it as it is because the spectral part also resynths the harmonic partials, so the "pure" tone of the guitar would be amplified, and the "distortion" sound would be less in volume).

This was the key of the question here... the spec also resynths the harmonic parts", and when we sum them with the additive, the distorted part appears "less" in volume...

Hug, I expect I'm not confusing you too much with this...

Peace,
Mau.

And thanks Andy for the patience!!! :S

Hi,

Thanks for the additional explanation ... I wasn't 100% sure about your original post, so took a guess that you were a bit confused about how it works in general, but I see what you mean now.

The best results form additive analysis will be obtained from "pure" tones, where the fundamental and harmonics of the original sample do not deviate too much from fixed values. Of course, very few sounds match this description in real life, but the point is that a clear sample of just a single note from an instrument will generally be reproduced more accurately than a sound with wild pitch modulation and lots of effects (for example).

With a distorted guitar the sample will have a fairly obvious fundamental with increasingly difficult to determine pitch information as we move up the frequency spectrum, and particularly when analysing chords. Although the distortion is "harmonic" i.e. it emphasises certain frequencies in tune with the original sound, there are also significant fluctuations in amplitude to deal with - which is (obviously) why it sounds distorted in the first place.

Based on this, I would guess the point at which the pitch information is more difficult to determine happens much lower in the overall frequency range with distorted sounds compared to (say) a single note of a flute. This results in the spectral component having the high pass filter cutoff set much lower than with other sounds. I don't think the spectral component of the sound is being used to reproduce the distortion, by the way ... it is more related to the algorithm struggling to identify useful pitch information - but I will find out more information about that.

Testing add + spec analysis of some random distorted guitar samples, the additive component is quite smooth and basically sounds a lot like the original guitar tone might, without any distortion applied. Depending on the original sample and import settings, the high pass of the spectral component is set to around 3500 Hz. This is adding a lot of artefacts and this does not sound very convincing as "fuzz".

So, I think the results are better using additive analysis only. The settings I used are "Single Octave" and "Best Frequency". Changing the waveform used for resynthesis also helps e.g. Sin Power 4 is a good one, or some of the square waves can also work well. I then used a source level low pass filter to reduce the high frequencies quite significantly, with a high resonance setting. This was followed by a Tube filter with fairly extreme settings. Both filters help to put the volume and power back in to the sound. Because the sample is slowed down, the attack sounds quite weak so I adjusted the start position to put the attack back in to the sound.

The results are not a perfect replica of the original sample, but do create an excellent basis for a new distorted guitar like sound. But ... I guess this is the long way of saying we are back where you first started ...

I'm going to ask Ben and John for more information about this and to check that my posts are OK (since they are the experts ...), but I think the actual answer is that this is one area where Alchemy v2 will make a significant improvements in terms of resynthesis quality.

Peace,
Andy.

I think I've gotta demo Alchemy's synth engine.
Sounds interesting

Hi again,

I spoke to Ben about my answers and have a few minor corrections to make ...

1) The difficulty of analysing the upper frequencies in a sample is because most sounds have a low ratio of partials to noise in the high frequencies. In other words, the musically useful information (frequency) is difficult to detect due to the noise floor of the recording.

2) The high pass filter used on the spectral part of add + spec resynthesis is something we will try to improve with Alchemy v2. At the moment Alchemy is not subtracting enough of the pitched component of the original sample.

3) Additive should be able to analyse sounds that vary in pitch, but as stated above it does work better with clear, single note sounds rather than sounds with lots of effects, or chords.

4) Distortion does not always occur at harmonic frequencies, which makes it difficult to extract the information from a sample, but again this is something we will do our best to improve in Alchemy v2.

Peace,
Andy.

Thanks soooo much Andy!
Again, If I want a real guitar... Iy just play it .
If I can find what I did like six months ago, you will be so proud of Alchemy and yourselfs...
I mean, I imported a Power Chord (A and E for example, you know) into additive only (I think) and it's was like... amazing! Then tweaking the pitch knob, etc... you know better than anyone what Alchemy is . And then formant filters for example, this I didn't do... but a very cool wah wah there (specially if you use band-pass, which are wider than formants and let more frequencies pass, having a vocal "flavour" but letting in more of the instruments own timbre).
Regarding the spectral part... maybe filter out not like with a high pass but the harmonic partials (bins, in this case ). I REALLY don't know how to do it, but maybe spectral should "look" at the additive engine after resynth itself and "take out" in every "snapshot" the partials that are in the additive part(I mean partials and not harmonics even in additive thinking mode because sometimes they are a bit pitch-changed in the additive engine, which is the real power of additive resynth in my opinion, and spectral is very sensitive to frequencies becaue every bin is a fixed freq)... Just an idea, don't know if it is possible... I know, or think, the whole idea of spectral was like adding the extra noisy sounds that are found in natural instrument, hammer strikes, human mouth blowing on a flute (that doesn't sound very good, sorry ). But maybe it shouldn't be a high-pass but a very specific "surgery" filter, more like eq. Like, some things go to additive, and the others to spectral. And in spectral you have noise-resynth mode (one of my favourites , kudos!) which should be awesome in this case, but, i'm just doing this like on my head... Sorry to write so much things that may or may be not useless... when I have more time to open Alchemy and re-do all of this experiments i'll come back with more "from my ears" ideas and impressions...

Oh, and I found something weird... I triggered a kick sample in sample mode... I lit off the trigger button on the ahdsr of the amp because somehow it eats the very first attack portion (upper freqs of the kick), and by doing this, it doesn't... but... In Sonar X1, every time I heat the play of Sonar, it takes like a few milliseconds to "start... if I hit start on the kick, the attack sounds "eaten", but if I go to the middle, like putting the cursor on the third 1/16th, the first kick (that it is the second of the loop actually) sounds just fine. So I think there is like a delay from where you hit play and when alchemy starts, at least with samples... every other kick sounds fantastic, even when it starts the loop again... so I pretty confident it is something between the play button and the Sonar starting to act and Alchemy. Maybe you want to try this out...
Again Andy, a pleasure "talking" to you . And thanks, really, for taking all the time to answer my long and twisted and overthought questions.
Oh, and I want a Camel Audio and or Alchemy, or one with both, T-Shirts . How can we do that?

Love, Insight, and Psychedelic Trance !
Mau...

Oh, I took a second read on your last post. I'm not talking abut importing a riff... Single note (or in this particular case, a power chord, simple, punky, not progressive metal with four note, like fifht in the 6th strings, fundamental on the 5th string and then the octaves of this too in 4th and 3rd strings, just 1st and 5st).

regarding noise to "actual sound" ratio, I know what you mean :S , Distorted guitars are a pain in the *** for this. Try guitar rig... you have to use too much gate, and then you can't play, because you have no dynamics to play with.
If it does that much of a sound when you don't even touch the guitar, or even mute the strings... can't imagine what it does when you're playing, although you may not hear it because what you play is louder... but it's there anyway.

But I will try to import a power chord (sampled by another person, not from my guitar) and give it a try again... Will send you the results... I mean, I don't know if you actually need that distortion and all that... only the additive does something really cool... maybe it's working bad, and not how it is intended too.. or it was in that particular moment... but it was so cool. Anyway, it will continue, I hope...

Mau.

nix808: It's far more than interesting . Trust me on this one. I say lots of stupid things... this is one of the very little exceptions .
Wish you best.

Mau