Author |
Message |
Tim Kleinert
Joined: Mar 12, 2004 Posts: 1148 Location: Zürich, Switzerland
Audio files: 7
G2 patch files: 236
|
Posted: Sat Feb 21, 2009 7:08 am Post subject:
PolyGranuSyn X |
 |
|
EDIT 30th April 2010: This design has been superceded. Improved version here:
http://electro-music.com/forum/topic-41067.html
hi folks,
As a sort of reconciliation with the G2 ( ), here's the final version of the series of "PolyGranuSyn" -patches.
Looking back at the history of these patches, PolyGranuSyn4 was the last design based on a true granular engine (4 grains per voice), the predecessors being preliminary versions as I was perfecting the techniques. PolyGranuSyn5 then was a different approach, recognising and extracting wavecycles, pitching them, clustering them and redistributing them along the time axis. Strictly speaking, it was incorrect to call it granular, as it doesn't use a grain-approach, but rather something more akin to time stretching like in old samplers, but in realtime. I first intended to improve on this second approach, having ideas to include "forcing to desired pitch" and formant control. But when building it, I noticed glitches and a certain "lo-fi-ness", that came along with the design and were unfortunately impossible to cure. (I might post it in the experimental section anyway.) "Lo-fi-ness" is something that all these patches had in common anyway, and I was getting sick of it, as -for my personal taste- they aren't really usable in a professional recording session. The granular PolyGranuSyn patches suffered from the whole readout/pitching-engine clocking at control rate, audibly degrading the signal. I got some requests to update this approach with audiorate techniques -and that is what I have done here. This is a high quality patch, really usable.
This means that everything is based on klutzy low level stuff -there are 7 DIY oscillator/counter circuits per voice in there. One problem with this stuff is always the keyboard tracking. I had to invent an "embedded DIY osc" per grain (core osc inside a pointer osc) in order to get the delay readouts to access the whole delay lines and still track over the musical range properly. But they do now -they track like a bloodhound ! The other, and real biggie, is that there is no easy way to patch an audiorate modulo circuit -the neat trick with the NoteScaler doensn't work, as it stays "blue" whatever you patch into it. (Modulos are crucial for this patch, there are 8 of them per voice. )So you have to go the klutzy way, eating up mem% like hell. As a consequence, this patch only features two grains instead of four. I've found out that this isn't much of a drawback. Sometimes it even sounds better, as there are less phase cancellations.
Other small glitches, like artefacts with fast attack times, have also been eliminated.
I saved some mem% by putting the sampling circuits into the FX area and piping stuff back into the voice area via 3 interslot busses. This can cause unwanted behaviour if other patches also use these. You'll probably want to use it standalone anyway, for polyphony reasons.
Enough blabla, here are the controls:
Input Section:
INPUT Source, Pad -select external input channel and pad
INPUT Monitor -listen to the input (running through the input limiter, drive and EQ)
SAMPLE Capture -samples audio (2.7 seconds)
INPUT Limiter -input dynamics
INPUT Drive -overdrive for the input, can be useful
INPUT EQ -EQ for the input, can be useful to add eg. "air" to voices
Playback Section:
SAMPLE Speed -playback speed of the sample (independent of pitch), positive and negative; 32 equals roughly original speed; assigned to the pitch stick in order to "cruise" through the sample
SPEED KBT -lets speed track the keyboard exponentially; useful for "pseudo sample playback" and also for audiorate grain clock stuff (see below)
SAMPLE KBTrig -restarts the sample at the set position at keypress; means that you can also have freeruning sample playback (nice for ambient stuff)
SAMPLE Position -playback starting position; can also be "scrubbed" during playback, hence assigned to modwheel.
POSITION Random -granular randomisation of the position
PITCH Semi, Fine, KBT, Random -grain pitch controls
TUNING TOOL -a monophonic reference pitch for tuning; makes audio bypass the FX section.
GRAIN CLOCK Coarse, Fine, KBT, Random -grain clock controls
G-MODE -toggle between audiorate and LFO grain clock; the latter is great for slow ambient stuff, the former has other interesting apps (see below)
GRAINS Overlap -controls the amount of grain crossfading overlap. High amounts are great with slow grain rates and some pitch randomisation, as it creates subtle beating like detuned oscillators.
Env Section:
VOLUME CONTOUR, VOICE Level -amplitude control
Filter Section:
HPF -a 24dB highpass filter; cannot be taken out of the audio path as it also serves for removing DC which can occur at some settings.
MULTIMODE FILTER -for further treatment of the sound
FX section:
FX Input Pad -for if you experience overloads...
EQ -self explanatory
CHORUS -self explanatory
DELAY -true stereo delay with gap control between the R and L delay times. Also features damping control which "tilts" the damping between allpass and lowpass response.
REVERB ...
OUTPUT ...
As you can see, there are no modulation generators (LFOs, ENVs) anymore in the patch. Personally, I've never used them -manual tweaking is way more fun. I consider it a bit pointless anyway to implement too much programmability into a patch that is ultimately based on volatile audio. Hence also the variations only demonstrate very basic stuff.
Variations:
1 -basic sample playback (a bit glitchy as the grain clock subtly bleeds into the audio)
2 -slow playback
3 -slow grain clock, some pitch random, FX (great for pads)
4 -audiorate grain clock, FX
5 -random warble
Programming tip:
There are basically two ways in which this patch operates:
1) At slow grain clock rates, pitch is determined by the grain pitch settings, and playback speed moves through the sample. This is great if you eg. want to turn a tonal sample (strings, your voice) into a nice pad sound (see mp3 example). Higher settings of grain overlap and subtle pitch randomisation work well here.
2) At audiorate grain clock rates however, the grain clock determines the pitch. Think of this like a osc hardsync scenario, where the grain clock is the driver osc and the grains are the slaved osc. This means that the grain pitch control actually turns into a nice formant control. Also, playback speed turns into a low rate phase modulation. This means that playback speed will also affect the resulting pitch -which is also a reason why playback speed features keyboard tracking (otherwise everything would be out of tune due to a linear pitch offset). So, if you encounter tracking problems with this approach, make sure speed KBT is active. Also, this sounds better with minimal grain overlap (imho).
I also included a mp3 demonstrating how variation 3 turns my lowly voice, singing "aahh", into a nice airy "rompler style" choir pad (grabbed the grain clock control at the end ) The vastly improved sound quality is especially noticable in the high frequencies.
7 voices for expanded G2s, 3 for unexpanded (sorry...)
Remember: this patch uses 3 interslot busses!
best,
tim
(...who now probably holds the record for e-m's longest post )
Description: |
Final version of the series of PolyGranuSyn-patches. All-audiorate algorithm for superior sound quality. |
|
 Download |
Filename: |
PolyGranuSynX_TK.pch2 |
Filesize: |
9.27 KB |
Downloaded: |
1973 Time(s) |
Description: |
Demo of PolyGranuSynX, using my voice and variation 3. |
|
 Download |
Filename: |
TimChoir.mp3 |
Filesize: |
1.27 MB |
Downloaded: |
1503 Time(s) |
Last edited by Tim Kleinert on Sat Sep 15, 2012 12:30 pm; edited 2 times in total |
|
Back to top
|
|
 |
dorremifasol

Joined: Sep 28, 2006 Posts: 814 Location: Barcelona, Spain
Audio files: 7
G2 patch files: 49
|
Posted: Sat Feb 21, 2009 11:37 am Post subject:
|
 |
|
Wow, this patch is quite amazing!!!  _________________ Cheers,
Albert |
|
Back to top
|
|
 |
mother misty

Joined: May 13, 2004 Posts: 681 Location: Ghent / Belgium
Audio files: 82
G2 patch files: 130
|
Posted: Sat Feb 21, 2009 2:46 pm Post subject:
|
 |
|
Stunning work tim, brilliant patching!
This will certainly become one of my favourite patches ever!
greets,
misty. _________________ http://www.myspace.com/mariodemeyer |
|
Back to top
|
|
 |
iPassenger

Joined: Jan 27, 2007 Posts: 1067 Location: Sheffield, UK
Audio files: 5
G2 patch files: 78
|
|
Back to top
|
|
 |
Fozzie

Joined: Jun 04, 2004 Posts: 875 Location: Near Wageningen, the Netherlands
Audio files: 8
G2 patch files: 49
|
Posted: Sun Feb 22, 2009 4:48 am Post subject:
|
 |
|
Superb! What a patch! The demo is so very un-G2-ish (if that exists anyway) in a very good way  _________________ Spinning at ~0.0000115740740741 Hz |
|
Back to top
|
|
 |
Christophe Duquesne

Joined: Jan 06, 2007 Posts: 17 Location: France
Audio files: 1
G2 patch files: 6
|
Posted: Sun Feb 22, 2009 7:20 am Post subject:
|
 |
|
Wahoou ... great patch....
I didn't even thought that quality level was available in a G2 for granular/sampling synthesis ....
That's going to give some more work to my G2 (if possible )....Many thanks _________________ Christophe Duquesne |
|
Back to top
|
|
 |
Chrono

Joined: May 18, 2008 Posts: 229 Location: NL
Audio files: 37
G2 patch files: 38
|
Posted: Sun Feb 22, 2009 6:21 pm Post subject:
|
 |
|
í havent connected my g2 yet after my live performance. but will do soon to test this out.
the mp3 sounds great! |
|
Back to top
|
|
 |
Chrono

Joined: May 18, 2008 Posts: 229 Location: NL
Audio files: 37
G2 patch files: 38
|
Posted: Mon Feb 23, 2009 9:55 am Post subject:
|
 |
|
Verrry Nice Tim!
diddnt know i had such a great voice  |
|
Back to top
|
|
 |
iPassenger

Joined: Jan 27, 2007 Posts: 1067 Location: Sheffield, UK
Audio files: 5
G2 patch files: 78
|
Posted: Mon Feb 23, 2009 11:11 am Post subject:
|
 |
|
Brilliant patch, had some truly unique sounds out of it on my first mess.
However, I may just be being rubbish (or wanting too much) but have you got any tips for getting a smoother loop when using the lfo grain control mode?
Cheers
R. _________________ iP (Ross)
- http://ipassenger.bandcamp.com
- http://soundcloud.com/ipassenger |
|
Back to top
|
|
 |
Chrono

Joined: May 18, 2008 Posts: 229 Location: NL
Audio files: 37
G2 patch files: 38
|
Posted: Mon Feb 23, 2009 3:55 pm Post subject:
|
 |
|
if im not wrong the LFO does not grain but only pitch. i asume OSC is Grain mode.
well anyways i defo hear my self beeing more false with LFO mode i cant sing and have a terrible voice unless i shout loud
Tim is it possible if i CAPTURE and speak a short sentence , then press a chord that all notes i play, playback the recording with the same length?
im not trying to whine or something , atm im recording my vocals through your patch for my solo performance in the Hemkade Saturday. Fresh food for my mpc1000 Last edited by Chrono on Mon Feb 23, 2009 4:30 pm; edited 1 time in total |
|
Back to top
|
|
 |
Tim Kleinert
Joined: Mar 12, 2004 Posts: 1148 Location: Zürich, Switzerland
Audio files: 7
G2 patch files: 236
|
Posted: Mon Feb 23, 2009 4:14 pm Post subject:
|
 |
|
iPassenger wrote: |
However, I may just be being rubbish (or wanting too much) but have you got any tips for getting a smoother loop when using the lfo grain control mode? |
The problem is that if the grain size is larger than the 2.7 second delay line, or the position in the sample is close to the end of the delay line, the playback will play over the end boundary of the delay line and wrap around to the beginning again, potentially causing a glitch. I could solve this by punching out the sample capturing perhaps 100ms or so before hitting the end boundary with a corresponding fade-out. This means however that there will be a very short gap of silence at that point. I actually did patch this, but found this gap of silence no less annoying than the potential glitch -so I ommited it again. I can throw it back in if you want. Just takes one module more.  |
|
Back to top
|
|
 |
Chrono

Joined: May 18, 2008 Posts: 229 Location: NL
Audio files: 37
G2 patch files: 38
|
Posted: Mon Feb 23, 2009 4:44 pm Post subject:
|
 |
|
2.7 is enough for my hardcore vocals , ill record them in parts. Ill post some demo's  |
|
Back to top
|
|
 |
iPassenger

Joined: Jan 27, 2007 Posts: 1067 Location: Sheffield, UK
Audio files: 5
G2 patch files: 78
|
|
Back to top
|
|
 |
Tim Kleinert
Joined: Mar 12, 2004 Posts: 1148 Location: Zürich, Switzerland
Audio files: 7
G2 patch files: 236
|
Posted: Wed Feb 25, 2009 4:15 am Post subject:
|
 |
|
The grain size is proportional to the grain clock speed. Higher speed, smaller grains.  |
|
Back to top
|
|
 |
Tim Kleinert
Joined: Mar 12, 2004 Posts: 1148 Location: Zürich, Switzerland
Audio files: 7
G2 patch files: 236
|
Posted: Wed Feb 25, 2009 4:17 am Post subject:
|
 |
|
Chrono wrote: |
Tim is it possible if i CAPTURE and speak a short sentence , then press a chord that all notes i play, playback the recording with the same length?
|
Yes of course, just switch off "speed KBT", so that the playback speed doesn't track the keyboard anymore. Then all voices will have the same speed, regardless of pitch. |
|
Back to top
|
|
 |
iPassenger

Joined: Jan 27, 2007 Posts: 1067 Location: Sheffield, UK
Audio files: 5
G2 patch files: 78
|
|
Back to top
|
|
 |
Tim Kleinert
Joined: Mar 12, 2004 Posts: 1148 Location: Zürich, Switzerland
Audio files: 7
G2 patch files: 236
|
Posted: Wed Feb 25, 2009 5:07 am Post subject:
|
 |
|
The wraparound glitch will happen regardless of grain size. The reason for this is that the capturing process samples the full 2.7 seconds of audio and then crossfades out back at the beginning. So, at the beginning of the 2.7 sample, there is a small crossfade with the end. However, the precision of the delay line taps isn't 100%, so that wrapping around the readout pointer to the beginning will cause a small glitch in the audio. I cannot help this, it is not the fault of the patch itself.
The only way to prevent this is to punch out the capturing process before the end of the 2.7 seconds (with a smooth decay of course), so that there is a brief gap of silence at the end, so that the wraparound will only deal with "silent" samples and thus not glitch. I can add this, it only takes an extra module.
If you want to use the patch as a crossfading-sampler (like I did in the mp3) just make sure that the readouts can never hit the wraparound point -meaning: make sure grain size is smaller than 2.7 seconds, and that the readout position is at a point where position+grainsize will not exceed the 2.7 seconds either. Also, speed has to be zero -otherwise the playback will move towards the wraparound sooner or later anyway. For this application, I therefore always keep position and speed at zero, and grain clock with KBT on. I then tweak the grain size to were it sounds glitchless. The clock KBT will ensure that, even with faster playback due to pitching, the grains will never exceed the limit (as grain size will also be attenuated accordingly due to clock KBT).
It's a bit tricky to understand how all these components interact.
Interestingly, the glitch doesn't seem so noticeable at higher grain rates. |
|
Back to top
|
|
 |
iPassenger

Joined: Jan 27, 2007 Posts: 1067 Location: Sheffield, UK
Audio files: 5
G2 patch files: 78
|
Posted: Wed Feb 25, 2009 6:19 am Post subject:
|
 |
|
tim wrote: |
It's a bit tricky to understand how all these components interact.
|
LOL, yr not kidding.
OK, I think I understand what you mean.. I'll give it a go when i get in.
Re:higher grain rates:
It isn't noticeable at all when the grain rate is set to audio, IIRC.
Re: the punch out extra module :
Do you mean that you use a sharp volume envelope applied to the sample to create a zero crossing at the loop point, effectively inserting a silent point? Presumably if this was fast enough, the silence wouldn't be that noticeable? Could you stick that version up as well for reference?
Unfortunately I have packed half of my studio up as I am getting ready to move house (groan) but I can easily get the mic out for more experiments. _________________ iP (Ross)
- http://ipassenger.bandcamp.com
- http://soundcloud.com/ipassenger |
|
Back to top
|
|
 |
iPassenger

Joined: Jan 27, 2007 Posts: 1067 Location: Sheffield, UK
Audio files: 5
G2 patch files: 78
|
Posted: Wed Feb 25, 2009 2:36 pm Post subject:
|
 |
|
Cheers for the tips, it is really working for me now. Got some great stuff just routing a filter swept osc mix (from another slot) and then moving and stuttering around the sweep, once sampled. _________________ iP (Ross)
- http://ipassenger.bandcamp.com
- http://soundcloud.com/ipassenger |
|
Back to top
|
|
 |
Neon Sega
Joined: Mar 31, 2006 Posts: 78 Location: ru msc
|
Posted: Wed Feb 25, 2009 4:23 pm Post subject:
|
 |
|
it's very amazing actually, sounds great indeed, unlike most G2 stuff |
|
Back to top
|
|
 |
lifeonmars
Joined: Dec 21, 2007 Posts: 23 Location: italy
Audio files: 1
|
Posted: Fri Mar 06, 2009 2:06 am Post subject:
|
 |
|
thank you! You are a genious! |
|
Back to top
|
|
 |
Chrono

Joined: May 18, 2008 Posts: 229 Location: NL
Audio files: 37
G2 patch files: 38
|
Posted: Sun Mar 08, 2009 12:24 pm Post subject:
|
 |
|
tim wrote: | Chrono wrote: |
Tim is it possible if i CAPTURE and speak a short sentence , then press a chord that all notes i play, playback the recording with the same length?
|
Yes of course, just switch off "speed KBT", so that the playback speed doesn't track the keyboard anymore. Then all voices will have the same speed, regardless of pitch. |
great ill have another go soon its truly sick patching, keep it up! |
|
Back to top
|
|
 |
bassfucker
Joined: Jul 05, 2009 Posts: 2 Location: berlin
|
Posted: Sun Jul 05, 2009 5:24 am Post subject:
Re: PolyGranuSyn X Subject description: Final version of the series of PolyGranuSyn-patches. All-audiorate algorithm for superior sound qual |
 |
|
wow, this sounds really great, I would like to use it !!!
But I can get nothing out of that patch ... it's not the easiest to understand what's going on there ...
Could you give s.th. like a step by step advice, how I can get something out of this great patch ?
That would be really great
cheers,
oliver |
|
Back to top
|
|
 |
Tim Kleinert
Joined: Mar 12, 2004 Posts: 1148 Location: Zürich, Switzerland
Audio files: 7
G2 patch files: 236
|
Posted: Sun Jul 05, 2009 5:41 am Post subject:
Re: PolyGranuSyn X Subject description: Final version of the series of PolyGranuSyn-patches. All-audiorate algorithm for superior sound qual |
 |
|
bassfucker wrote: | wow, this sounds really great, I would like to use it !!!
But I can get nothing out of that patch ... it's not the easiest to understand what's going on there ...
Could you give s.th. like a step by step advice, how I can get something out of this great patch ?
That would be really great
cheers,
oliver |
Hi Oliver,
This is a granular synthesis patch. Granular synthesis is based on an audio sample. The G2 however cannot store or play back samples by anyway. I developed a technique that can "abuse" a 2,7 seconds delay line as a volatile sampling memory. So, in order to hear anything, you first need to sample some audio into the delay line. The "INPUT" section defines the audio input from which you want to sample some audio. Eg., select channel 1, and plug a sound source of your choice (another synth, CD player, microphone in the mic input, whatever) into it. When you press the "Capture" button, 2,7 seconds of incoming audio will be sampled into the delay line. Now you can start to process this with the granular algorithm. In the demonstration mp3, I just sung "aah" into the microphone, captured it, set parameters to my liking and then played some chords on the keyboard.
Remember that the sampled audio can unfortunately not be stored by anyway. So if you change patches, it will be gone, and you will have to capture some audio again. Best is to capture what you want, experiment and tweak the parameters to your liking (and/or use the assigned performance controllers) and record it.
cheers,
tim |
|
Back to top
|
|
 |
bassfucker
Joined: Jul 05, 2009 Posts: 2 Location: berlin
|
Posted: Sun Jul 05, 2009 5:48 am Post subject:
|
 |
|
ok,
got it to work now :
1. press capture Button
2. Play any Key
Pretty easy when you know that.
Thanks for the fast answer and thanks for the great patch !
Oli |
|
Back to top
|
|
 |
|