so..Resynthesis

VST, AU, AAX, etc. plug-in Virtual Instruments discussion
User avatar
ChamomileShark
KVRAF
1829 posts since 12 May, 2004 from Oxford, UK

Post Mon Jun 29, 2020 12:12 pm

I seem to remember this was the holy grail back in the 70s. It doesn't seem to be that common these days (Synclavier 2V, Alechemy 2?). Is that because the technology is called something else these days? Is it just a kind of wavetable were frames get repeated or granular?

And what apart from Synclavier 2 and Alchemy (was it in 1.5?) have it?

neverbeeninariot
KVRist
346 posts since 3 Feb, 2015 from UK

Re: so..Resynthesis

Post Mon Jun 29, 2020 1:22 pm

IL Harmor - https://www.youtube.com/watch?v=sOd8FutW-nk (seamless talks at 100 kph, be warned). The resynthesis is impressive though.

User avatar
VariKusBrainZ
KVRAF
9121 posts since 16 Dec, 2002

Re: so..Resynthesis

Post Mon Jun 29, 2020 3:05 pm

DiscoDSP Vertigo
IL Morphine
Virsyn Cube


The list goes on.
There was a surge of releases about 10 years ago.

xoxos
Mr Entertainment
12179 posts since 30 Apr, 2002 from i might peeramid

Re: so..Resynthesis

Post Mon Jun 29, 2020 3:44 pm

xoxos resyn2 from 2019, when i looked there were a few free instruments that used phase vocoding but limited features and implementation so i mak. it can do some clean things but i am no expert on programming for memory handling.
you come and go, you come and go. amitabha neither a follower nor a leader be tagore "where roads are made i lose my way" where there is certainty, consideration is absent.

kavin2134
KVRist
77 posts since 27 Mar, 2020

Re: so..Resynthesis

Post Tue Jun 30, 2020 4:56 am

Well, Parawave Rapid also has the resynthesise feature. Along with that, it should get granular pretty soon. It is a beast synth. Virtual analog, wavetables, samples, ROM, Granular, resynthesise and a little FM all available to use in 8 layers. Do check it out.

Kwurqx
KVRian
576 posts since 15 Jun, 2017

Re: so..Resynthesis

Post Tue Jun 30, 2020 6:37 am

Resynthesis comes in many forms. Generally it's trying to analyse a sound and map it to some synthesis model.

E.g. by doing an FFT analysis on a sound or sample and then reproduce the sound by iFFT of the resulting spectrum. That would map it to an Additive synthesis model.

Maybe not even do the FFT and using Wave Tables (based on analysis/recordings of a sound). Might that be called resynthesis? Since it reconstructs a sound without actual replaying the full sample/recording.

The Hartmann Neuron was an attempt to move away from "classic" synthesis techniques, towards analysis and resynthesis of various properties of sounds.
https://en.wikipedia.org/wiki/Hartmann_Neuron
http://www.neuronsynth.com/html/Home.html

User avatar
ChamomileShark
KVRAF
1829 posts since 12 May, 2004 from Oxford, UK

Re: so..Resynthesis

Post Tue Jun 30, 2020 6:43 am

xoxos wrote:
Mon Jun 29, 2020 3:44 pm
xoxos resyn2 from 2019, when i looked there were a few free instruments that used phase vocoding but limited features and implementation so i mak. it can do some clean things but i am no expert on programming for memory handling.
I think I'll give that a try, thanks!

User avatar
ChamomileShark
KVRAF
1829 posts since 12 May, 2004 from Oxford, UK

Re: so..Resynthesis

Post Tue Jun 30, 2020 6:45 am

Kwurqx, I was thinking the same. Iget the feeling that resynthesis has moved on and now there are other techniques to do what would have been called resynthesis back in the day.

User avatar
pdxindy
KVRAF
16855 posts since 3 Feb, 2005 from in the wilds

Re: so..Resynthesis

Post Tue Jun 30, 2020 7:12 am

PPG Infinite Pro

SneakyBeats
KVRer
19 posts since 29 Jun, 2020

Re: so..Resynthesis

Post Tue Jun 30, 2020 7:43 am

Diversion?

User avatar
RichN
KVRist
229 posts since 4 Dec, 2004 from The Netherlands

Re: so..Resynthesis

Post Tue Jun 30, 2020 8:46 am

Steinberg Backbone

User avatar
Dirtgrain
KVRian
1032 posts since 12 Jan, 2019

Re: so..Resynthesis

Post Tue Jun 30, 2020 8:47 am

pdxindy wrote:
Tue Jun 30, 2020 7:12 am
PPG Infinite Pro
Unobtainable. :(
"When the going gets weird, the weird turn pro" - Hunter S. Thompson

User avatar
Dirtgrain
KVRian
1032 posts since 12 Jan, 2019

Re: so..Resynthesis

Post Tue Jun 30, 2020 8:51 am

SeamlessR in the video says what Harmor is not creating wave tables (like Halion 6 does). But as he scrubs through sound (sound image, whatever you call it), at each moment it is just like a cycle of a wave table, right? It's just not handled like wave tables somehow?
"When the going gets weird, the weird turn pro" - Hunter S. Thompson

feralsystems
KVRist
78 posts since 7 Nov, 2012 from Philadelphia

Re: so..Resynthesis

Post Tue Jun 30, 2020 8:56 am

Not exactly resynthesis, but these folks use neural networks to analyze audio and create new content based on that analysis.

https://dadabots.com/

cron
KVRAF
3353 posts since 27 Dec, 2002 from North East England

Re: so..Resynthesis

Post Tue Jun 30, 2020 9:41 am

Dirtgrain wrote:
Tue Jun 30, 2020 8:51 am
SeamlessR in the video says what Harmor is not creating wave tables (like Halion 6 does). But as he scrubs through sound (sound image, whatever you call it), at each moment it is just like a cycle of a wave table, right? It's just not handled like wave tables somehow?
A developer is probably best placed to answer why this is different to wavetable synthesis, so take the following with a pinch of salt.

A limitation of wavetable (re)synthesis is that you can't capture/reproduce inharmonic overtones. Look at your wavetable oscillator under a spectrum analyser and all the energy in there will be at multiples of the fundamental frequency. Some wavetable synths implement additional functions with the potential to produce inharmonic tones (e.g. FMing one oscillator with another), but a 'raw' wavetable or any 'oscillator effect' that operates directly on the wavetable will never produce anything inharmonic.

Think about how oscillator sync works in a VA - no matter what the pitch of the synced oscillator is, the cycle being reset each time the master oscillator completes a cycle means it's always 'locked' to the master oscillator frequency. The same thing happens with the audio content in any wavetable. You're basically looping the wavetable frame at, for instance, 440 times per second for the A above middle C. You can consider this looping to be the master oscillator, while all the audio content inside the wavetable is 'synced' to it.

Harmor has no such limitation. There are a few methods of doing analysis/synthesis in the spectral domain, but Harmor seems to be driving a bank of sine waves. As he's scrubbing the sound, he's effectively telling the synth how many sine waves to play, what frequency they should be, and what amplitude they should be.

Return to “Instruments”