Additive synthesis anyone??

[briandc]

Hi everyone,
they say that in theory, you can make any sound desired if you add enough sine waves. (My first thought that comes to mind is an acoustic piano emulation.)

-Has anyone ever re-created a piano sound, using additive synthesis?

-Also, is anyone here really "into" additive synthesis? If so, got demos of what you've done with it?


brian

PS: these questions could also be posed to other synthesis types as well: anyone out there really into vector synthesis? or other types of synthesis?

[ItsMeOnly]

Hi everyone,
they say that in theory, you can make any sound desired if you add enough sine waves. (My first thought that comes to mind is an acoustic piano emulation.)

-Has anyone ever re-created a piano sound, using additive synthesis?

Kurzweil had

-Also, is anyone here really "into" additive synthesis? If so, got demos of what you've done with it?

http://ftp.do.id.uw.edu.pl/pub/music/Samples/choir2.wav
http://ftp.do.id.uw.edu.pl/pub/music/Sa ... titled.wav
http://ftp.do.id.uw.edu.pl/pub/music/K3 ... nowsMe.mp3

[Bitexion]

The Kawai K5000 synths are purely additive. I've heard some pretty good acoustic reacreated sounds from those.
Obviously h**l to program with thousands of parameters per patch but hey..it's additive.

[Hybrid88]

Thinking of that, they must have been incredibly brave or just plain nuts to make such a deep synth with an interface like that. I know the technology of the time was limited but still... similarly the DX7 I suppose, but we all know that was only famous because of its presets.

It's a shame no one has persisted with a hardware additive synth like the K5000 because with todays technology editing sounds could be made so much easier and intuitive.

[garranimal]

Had a K5000. Sounded phenomenal. The formant filters are bonkers. Learned how to program it in depth. But it was extremely tedious, so I flipped it.

The Iris software is pretty cool. It can analyze input audio, create a granular model, and alter harmonic spectra by doodling in a graphic representation. Iris is incompatible with my Win XP system

My demo of the K5000

[briandc]

@ItsMeOnly: thanks for the demos. Too bad they're so short... But good to hear!

@Bitexion: thanks for the recommendation! I found a Kawai K5000 pdf manual online-- looks very interesting indeed!

@garranimal: thanks for posting your video. Very nice! Are there lots of effects added to the sounds, or do they sound good on their own too?



brian

[Soul_Processor]

if you're a bit into mathematics, look up "fourier transformation" which is the mathematical process of combining sines and cosines to approximate other waveforms.

[briandc]

@Soul_Processor: Thanks, I'll look into that!

According to the Additive Synthesis page on Sound-On-Sound (http://www.soundonsound.com/sos/jun00/a ... nthsec.htm), additive synthesis is capable of more sound timbres than subtractive synthesis. (No news to some of you here, I'm sure.)

So why is it so rarely used? Because it takes more time to work with? Was no one really into buying synths that used this method? (Even FM isn't so easy to understand, and yet the DX7 was a hit!)


brian

[pflosi]

Well, if you go read up on Fourier, you will realize that additive can, theoretically, make any (periodic) sound - as long as you have enough sine oscillators with level control. Cannot really say that for FM or subtractive.

FM is extremely easy compared to additive. Four operators with, say, ten fixed algorithms and you already have a potentially infinite amount of sounds.

Additive, you need hundreds of sine waves. Literally. All with their own envelopes, etc. IIRC, the K5000 (which also features PCM, not strictly additive only as claimed above) has six layers of 64 sine operators (making up harmonic series). That's a lot of parameters compared to the ten seconds it takes to dial up a quick and simple sound on a DX7 (from the init voice). Of course, 99% of people just use(d) the DX7 presets anyways.

Besides that, the labels are kinda stupid anyways. There's FM in every synth ever, lots of addition going on in traditional "subtractive" synths, the K5000 has a subtractive filter, too, etc.

Cheers

[meatballfulton]

So why is it so rarely used? Because it takes more time to work with?

This.

Quick, describe how the harmonics of a piano change over time, including factoring in velocity, sustain pedal, etc. That's what you have to program.

Some things are easy to program like sine waves but if you want to emulate filter sweeps it's not quite so simple.

Best use of additive I can think of is the additive resynthesis in Alchemy (now part of Logic). Load a sample and the synth calculates an additive equivalent. It's not perfect but it works pretty well. You can also build up additive sounds from scratch but it's way easier to use a sample as the starting point.

Alchemy also has "spectral resynthesis" which can be used alone or alongside the additive, as seen in this video. How does it work? I dunno, just twist knobs and see what happens

[Baus]

I am into Transwave synthesis

Robin.

[briandc]

Well, if you go read up on Fourier, you will realize that additive can, theoretically, make any (periodic) sound - as long as you have enough sine oscillators with level control. Cannot really say that for FM or subtractive.

FM is extremely easy compared to additive. Four operators with, say, ten fixed algorithms and you already have a potentially infinite amount of sounds.

Additive, you need hundreds of sine waves. Literally. All with their own envelopes, etc. IIRC, the K5000 (which also features PCM, not strictly additive only as claimed above) has six layers of 64 sine operators (making up harmonic series). That's a lot of parameters compared to the ten seconds it takes to dial up a quick and simple sound on a DX7 (from the init voice). Of course, 99% of people just use(d) the DX7 presets anyways.

Besides that, the labels are kinda stupid anyways. There's FM in every synth ever, lots of addition going on in traditional "subtractive" synths, the K5000 has a subtractive filter, too, etc.

Cheers

Some good observations. And yes, labels kind of confuse the issue. Maybe there are no "truly unique" synthesis methods. Or maybe there are. It's as though we're dealing with two basic issues: how to contruct the basic tone (samples, complex waves, stacking sine waves, etc) and how to modulate it (filter it, freq mod, amp mod, cross-mod, etc). -- Isn't an LFO basically doing FM?


brian

[briandc]

So why is it so rarely used? Because it takes more time to work with?

This.

Quick, describe how the harmonics of a piano change over time, including factoring in velocity, sustain pedal, etc. That's what you have to program.

Some things are easy to program like sine waves but if you want to emulate filter sweeps it's not quite so simple.

Best use of additive I can think of is the additive resynthesis in Alchemy (now part of Logic). Load a sample and the synth calculates an additive equivalent. It's not perfect but it works pretty well. You can also build up additive sounds from scratch but it's way easier to use a sample as the starting point.

Alchemy also has "spectral resynthesis" which can be used alone or alongside the additive, as seen in this video. How does it work? I dunno, just twist knobs and see what happens

Thanks for the video! Then I got sidetracked looking for possible fourier calculators in my linux repositories. Found a few interesting apps that might interest (highlighted a few) :

libfftw3-3 - library for computing Fast Fourier Transforms
libfftw3-dbg - Library for computing Fast Fourier Transforms - debug symbols
libfftw3-dev - library for computing Fast Fourier Transforms
libfftw3-doc - documentation for fftw version 3
libgsl0ldbl - GNU Scientific Library (GSL) -- library package
python-numpy - Numerical Python adds a fast array facility to the Python language
python-numpy-dbg - Fast array facility to the Python language (debug extension)
python-numpy-doc - NumPy documentation
python3-numpy - Numerical Python adds a fast array facility to the Python language
python3-numpy-dbg - Fast array facility to the Python language (debug extension)
bugsx - program to evolve biomorphs using genetic algorithms
horae - interactive graphical processing and analysis of EXAFS data
cl-fftw3 - Common Lisp package for using the FFTW3 library
drawxtl - crystal structure viewer
enscribe - convert images into sounds
euler - interactive mathematical programming environment
fftw-dev - library for computing Fast Fourier Transforms
fftw-docs - documentation for fftw
fftw2 - library for computing Fast Fourier Transforms
glam2 - gapped protein motifs from unaligned sequences
gnucap - GNU Circuit Analysis package
kwave - sound editor for KDE
libalglib-2.6.0 - A cross-platform numerical analysis and data processing library
libalglib-2.6.0-dbg - Debugging symbols for the alglib library
libalglib-dev - Development files for the alglib library
libaudiomask-dev - Audio masking threshold estimation lib headers, docs and examples
libaudiomask1 - Audio masking threshold estimation library
libnewmat10-dev - matrix manipulations library (C++ headers files)
libnewmat10ldbl - matrix manipulations library (C++)
libsbsms-dev - Subband Sinusoidal Modeling Synthesis (development files)
libsbsms10 - Subband Sinusoidal Modeling Synthesis
libsphere-dev - Development files for Spherepack scientific library
sfftw-dev - library for computing Fast Fourier Transforms
sfftw2 - library for computing Fast Fourier Transforms
siggen - Waveform generation tools
slang-gsl - GNU Scientific Library binding for S-Lang
synaesthesia - A program for representing sounds visually
yorick-mira - optical intreferometry image reconstruction within Yorick
yorick-yeti-fftw - FFT plugin for the Yorick language
texlive-fonts-extra - TeX Live: Extra fonts
fftw3-static - library for computing Fast Fourier Transforms (static)
fftw2-static - library for computing Fast Fourier Transforms (static)


Anyone ever worked with these?


brian

[briandc]

I am into Transwave synthesis

Robin.

Is that the same thing as Transwave Wavetable Synthesis? (ie. Esoniq VFX?) Or do you use a software synth with Transwave Synthesis? What kinds of sounds do you use it for?


brian

[pflosi]

Isn't an LFO basically doing FM?

Yes, that's one of the reasons why I wrote that there's FM in every synth ever.

Heck, you don't even need to go that far. Playing the keys or sequencing a synth is FM, too.

Then I got sidetracked looking for possible fourier calculators in my linux repositories. Found a few interesting apps that might interest (highlighted a few) :

enscribe - convert images into sounds

Aphex Twin famously used that technique

Aphex Face in Windowlicker:



Spiral at end of same track:



Venetian Snares - Look (Songs about my Cats):



Cheers!