electro-music.com   Dedicated to experimental electro-acoustic
and electronic music
 
    Front Page  |  Articles  |  Radio
 |  Media  |  Forum  |  Wiki  |  Links  |  Store
Forum with support of Syndicator RSS
 FAQFAQ   CalendarCalendar   SearchSearch   MemberlistMemberlist   UsergroupsUsergroups   LinksLinks
 RegisterRegister   ProfileProfile   Log in to check your private messagesLog in to check your private messages   Log inLog in  Chat RoomChat Room 
 Forum index » Clavia Nord Modular » G2 Building Blocks
Suboctave generator plus analog warmth
Post new topic   Reply to topic Moderators: Nord Modular Editors
Page 1 of 1 [15 Posts]
View unread posts
View new posts in the last week
Mark the topic unread :: View previous topic :: View next topic
Author Message
Rob



Joined: Mar 29, 2004
Posts: 578
Location: The Hague/Netherlands/EC
G2 patch files: 109

PostPosted: Thu Mar 17, 2005 2:29 pm    Post subject: Suboctave generator plus analog warmth Reply with quote  Mark this post and the followings unread

In green is a suboctave generator that retains some of the sonic aspects of its input waveform. The idea is to gate every alternate cycle of the input waveform, and have control whether it passes through the gate in a controllable amount between normally or inverted.

This type of suboctave gives just some extra beef to a single oscillator model. The modulation is useful to e.g. give accents to some notes from a sequencer by faking an extra bass note, simulate an octave popping effect, modulate by an envelope, etc.

Extra in purple is a little 'analog warmth' tone control made with two allpass filters constructed by integrating mixers in a very simple but effective setup. Just one single knob allows to warm the sound in a controllable amount.

Note that the allpass filters tend to tilt the whole spectrum rather than work as hi and lo shelving, as they interact with each other on the original sound. Poles are set to roughly 250 Hz and 2kHz, so the spectrum is split into roughly three decades (ideally 20Hz <-> 200Hz <-> 2kHz <-> 20kHz) and are premixed in a 2:1 ratio, which works just fine when using '6dB' allpass filters.


SubOctGenerator.pch2
 Description:
SubOctave generator that retains some character of the input waveform (green) plus a simple analog warmth filter (purple)

Download
 Filename:  SubOctGenerator.pch2
 Filesize:  2.46 KB
 Downloaded:  1133 Time(s)

Back to top
View user's profile Send private message Send e-mail Visit poster's website
mosc
Site Admin


Joined: Jan 31, 2003
Posts: 17550
Location: Allentown, PA
Audio files: 123
G2 patch files: 60

PostPosted: Thu Mar 17, 2005 3:03 pm    Post subject: Reply with quote  Mark this post and the followings unread

Can you explain how you make all pass filters with mixers only? How do you know where the poles are? Is there some formula or somthing? To many of us, this is just magic. scratch
_________________
--Howard
my music and other stuff
Back to top
View user's profile Send private message Send e-mail Visit poster's website AIM Address
Blue Hell
Site Admin


Joined: Apr 03, 2004
Posts: 20273
Location: The Netherlands, Enschede
Audio files: 143
G2 patch files: 318

PostPosted: Thu Mar 17, 2005 3:04 pm    Post subject: Reply with quote  Mark this post and the followings unread

Thanks for posting the warmth filter Rob - dirt cheap (even in stereo) and just what was needed for the patch I'm working on.

I was just about to go out and explore your website for something like this :-)

Jan.
Back to top
View user's profile Send private message Visit poster's website
Rob



Joined: Mar 29, 2004
Posts: 578
Location: The Hague/Netherlands/EC
G2 patch files: 109

PostPosted: Thu Mar 17, 2005 5:09 pm    Post subject: Reply with quote  Mark this post and the followings unread

mosc wrote:
Can you explain how you make all pass filters with mixers only? How do you know where the poles are? Is there some formula or somthing? To many of us, this is just magic. scratch


Its quite simple actually. The output of a mixer acts as the 'Z-1 sample' to its inputs. So, if a feedback connection is made, the mixer turns into what is named an integrator. Meaning that if a sample goes into the signal input a part of it will be fed back and influence the next samples, 'smearing out' the input signal over time. Which is for all practical purposes exactly the same as a 6dB lowpass filter. One could also say that part of the energy that is in a single input sample will be stored in the integrator and be released in the next samples until its energy is virtually gone.

If feedback is increased, the input signal level to the signal input must be lowered in a way that the sum of the two input knobs is 100%. So, if the feedback level knob is at 75% the signal input knob must be at 25%, this to keep the signal at the same level or 'pass the input signal with unity gain'. The mixer should be in linear mode to keep things simple.

Tuning is simple as well, it is (100 minus feedback percentage) times (96kHz divided by two times pi = roughly 16kHz), which is roughly the set precentage of the input knob times 16kHz. So, if the knobs are set to 50/50 the cutoff is 50% of 16kHz = 8kHz, if set to 12.5/87.5 it is 12.5% of 16kHz = 2kHz. The lowest possible setting is 1.6/98/4 which yields roughly 250Hz. Of course there are exact formulas and there are problems when the cutoff is higher as roughly one/eigth of the sample rate, but that is not of any concern here.

An allpass is made by first creating a lowpass filter with one mixer (or a standard 6dB filter which tunes better but is a little more dsp expensive). Then a highpass is made by subtracting the output of the lowpass from the original signal. Adding the lowpass to the highpass will regenerate exactly the original signal again, provided there are no delays of a calculation sample, which is always ok on the G2 but hardly ever on the old NM. An allpass is created by not adding the lowpass and the highpass but subtracting the highpass from the lowpass. This will reverse the high in phase, but as there is a smooth slope there is also the smooth phase shift that characterises an allpass filter. All frequencies will be passed so for all practical purposes this is the closest as one can get to an allpass filter on the G2. So, what is needed to create an allpass is a lowpass filter and two mixer modules, thats all.
Then there is all sorts of stuff one can do with allpass filters. This filtering is one example, but inserting an allpass filter set to 1kHz to 4kHz in a feedback loop of an echo delay will vastly improve the naturalness of the echo. Basically allpass filters are very useful to model effects that happen when sound travels through air, reflect from walls, etc. Atmospheric effects, so to say.

Basically the effect of an allpass filter is dispersion, not unlike how a ray of white light disperses into a colourful spectrum when passing through a prisma. But considering the ear dispersion works out with different effects compared to the eye, as phase dispersion of the sound spectrum is only heard when the dispersed sound is mixed with other sounds or its own undispersed original. Just simply passing an audio signal through an allpass filter doesn't seem to do much at all. Of course we all know the phaser and its peculiar sonic effect, which is created by mixing the dispersed signal with the original signal. Then the effect of allpass filters suddenly becomes clear. In the atmosphere the effect seems less dramatic, but it is dramatically there as it adds a lot to the 'naturalness' of sound, the sonic 'colour' of the space one is in, etc. But it is more a subconsious hearing adding to feeling good in a room or feeling unnerved. Also the famous 'sounds good' and 'sounds lousy' have to do with the subtleties that are partly created by dispersion.

These days there are two types of allpass filters, one that has a smooth phaseshift, varying from between a few degrees for the low up to 90 or 180 degrees for the very high frequencies. The phase shift curve over the spectrum is not linear but more like an S shape. Additionally, longer delay lines are also considered allpass filters, as they can pass all frequencies and there will be considerable phase shifts between input and output of several times 360 degrees or several cycles of the waveform. With longer delay lines the phase shift curve is not like an S shape. The first type is very much like a single prism, but the delay line thingies are more like a row of prisms that project several colour spectra next to each other. This regrettably makes a lot of literature on allpass filters a little obscure, as sometimes it is not clear which type is meant. E.g. Curtis Roads refers basically to the delay line type in his 'Computer music turorial', which can put an unattentive reader on the wrong foot. The sonic difference caused by the two types is like the difference between the phaser module and the comb filter module, definitely distinctly different.

Mixing the outputs of two parallel allpass filters with the input signal creates an interesting type of tone control by 'bending' the spectrum. There is quite a lot of settings possible and the 12dB multimode filter in its '6dB BP' allpass mode can also be used. Attached is the tone control as in the previous patch, but in variation 2 the allpass pole frequencies are shifted up by one octave and the mix is 4:3. This tilts the spectrum in a less straight line with a very broad peak in the middle of the spectrum. You will hear that this gives the simple sawtooth sound a lot more 'presence'. This setting works out very well here, but might be exaggerated on other types of sound. The setting in variation 1 is the most neutral, where the spectrum is tilted in the most straight fashion (well, if I got my math right). Using two Multimode filters with optionally increased resonance will give extra coloration to the sound like in the other patch.


WarmthTrickSVF.pch2
 Description:
Tone filter using two parallel Multimode filters in allpass mode

Download
 Filename:  WarmthTrickSVF.pch2
 Filesize:  1.85 KB
 Downloaded:  983 Time(s)


WarmthTrick.pch2
 Description:
Basic tone filtering by using two parallel allpass filters

Download
 Filename:  WarmthTrick.pch2
 Filesize:  2.01 KB
 Downloaded:  994 Time(s)

Back to top
View user's profile Send private message Send e-mail Visit poster's website
The Why Project



Joined: Dec 05, 2004
Posts: 118
Location: Ireland
G2 patch files: 3

PostPosted: Thu Mar 17, 2005 5:38 pm    Post subject: Reply with quote  Mark this post and the followings unread

Rob,

Once again, your knowledge, but even more so the use of it is simply
amazing! Hail the Master
I sometimes feel that if I had only 10% of that knowledge, I would be
able to solve the world's biggest problems with the G2
Thank you for keeping to share it with the community.

props

This tutorial on allpass filters came as a godsend today... I'm working on
putting together a 'vintage' AGC processor with the G2, just to see how
it sounds... but also as a reaction on the recent compressor thread.
One of the building blocks my AGC is a 'symmetra peak', a device that
provides a better peak average between negative and positive parts of a
waveform. This is done by cascading 4 1st order allpass filters with their
poles on 200 Hz (for vocals that is)...

Regards,

The Why Project
Back to top
View user's profile Send private message Visit poster's website
mosc
Site Admin


Joined: Jan 31, 2003
Posts: 17550
Location: Allentown, PA
Audio files: 123
G2 patch files: 60

PostPosted: Thu Mar 17, 2005 6:54 pm    Post subject: Reply with quote  Mark this post and the followings unread

Thanks, Rob.

I played around with your patches and it's interesting to hear the effects of changing the mixer settings and polarity. This is cool stuff. When I look at the screen and see a bunch of mixers connected in strange ways with the knobs at certain numbers, it doesn't jump out as what it is - in this case a tone control. From that point of view I like the use of the MM filter for an all-pass filter, at least you can see the frequency (worth the DSP to me). It would be nice if Clavia would update the MM filter to some how indicate that you get all-pass with the 6 dB setting and band-pass with the 12 dB setting. Embarassed

By changing the frequencies of the filters and the coefficients and polarity on the mixer you can get lots of great tone jeffects, but it is hard to visualize what you are actually getting. If you don't do the math and just play with the knobs - this is called "art control". If you design the filter to have a certain characteristic and do the math to set the knobs, then this is "mind control". A new module with visual feedback like the eq modules have would be a great way to merge the two. Wink

_________________
--Howard
my music and other stuff
Back to top
View user's profile Send private message Send e-mail Visit poster's website AIM Address
Wan



Joined: Mar 31, 2004
Posts: 255
Location: Netherlands, Ugchelen
Audio files: 4
G2 patch files: 46

PostPosted: Fri Mar 18, 2005 2:59 am    Post subject: Reply with quote  Mark this post and the followings unread

Rob wrote:
Its quite simple actually....


Ah...

You are so capable to tell things that seem complex in a very understandable way. I love you (r explanations).

Thanks Exclamation .

_________________
Grtz Wan

Last edited by Wan on Fri Mar 18, 2005 5:27 am; edited 1 time in total
Back to top
View user's profile Send private message Visit poster's website
Rob



Joined: Mar 29, 2004
Posts: 578
Location: The Hague/Netherlands/EC
G2 patch files: 109

PostPosted: Fri Mar 18, 2005 4:19 am    Post subject: Reply with quote  Mark this post and the followings unread

The Why Project wrote:
Rob,

This tutorial on allpass filters came as a godsend today... I'm working on
putting together a 'vintage' AGC processor with the G2, just to see how
it sounds... but also as a reaction on the recent compressor thread.
One of the building blocks my AGC is a 'symmetra peak', a device that
provides a better peak average between negative and positive parts of a
waveform. This is done by cascading 4 1st order allpass filters with their
poles on 200 Hz (for vocals that is)...

Regards,

The Why Project


Yeah, realtime compressors are not at all that easy to construct, mainly because for ideal functioning they need information that can only be available in the near future.

In practice there are two things of concern:
1) There is always a short delay before a compressor gets into action. E.g. if a percussive hit is fed into a compressor the fast attack gets only slightly compressed. Simply because it takes a while before the compressor knows it needs to start to compress. Depending on the type of compressor this can pose problems. E.g. if the compressor takes a reference level and compares its own output signal to this reference level to see if it needs to attenuate or to amplify, the compressor is probably in amplify mode when the hit comes. As low sound level before the hit means amplification, so when the hit comes the amplifier is still on. This means that the attack of the hit will not be compressed, but instead be amplified. With nasty clipping as a result in a digital system. But it is this type of compressor that can do that deep compression that is popular today. It is almost impossible to get this right in realtime in the digital domain.
2) When a deep ratio realtime compressor is based on the idea of extracting a control value from the compressor output to control the gain of the compressor, there must be a feedback path. And always when there is a feedback path in a digital system there is the tendency to oscillate at half the sample rate. This tendency must be taken care of by bandlimiting the bandwidth. The point is this, a feedback loop is always a resonator like a resonant filter. If feedback gain is increased the resonance increases as well. Then when a strong signal hits the circuit it can excite this resonance and cause an oscillation to start. This means that there must always be high frequency damping in a feedback loop, which will lower the resonant frequency of the feedback circuit, which in practice means that much more energy is needed to get the circuit into selfoscillation. But this damping also means that the feedback control loop of the compressor needs to be damped and that leads to the mentioned point 1) time delay before the control signal starts to correct the gain.

The balance that needs to be found is no chance for nasty high frequency oscillations, which means a slower response, and a speed that is fast enough to prevent strong transient sounds to cause clipping.

An AGC circuit is often very slow, e.g. in radio it is not uncommon to use AGC circuits directly before the radio transmitter that have a time constant of sometimes a minute or more. But the only goal for such a compressor is to prevent spoiling of power in the radio transmitter by trying to establish a fixed modulation level of the carrier wave. The same goes for the AGC circuits in the cheap mono cassette players from the seventies, though these are a bit faster and compress and expand deeper.

Fast realtime compression with very high (over 1:1.5) compression rates is a nightmare in the digital domain. At the NM Event two years ago in London there was a nice presentation by 'Edward the Compressor', the guy behind JoeMeek. He claimed that best resuts are still with 'analog' cadmiumsulfide light cells, and he drives them by power leds that are 'put in overdrive' on the attack of a strong percussive hit. I can very well understand why.

Note that any allpass filtering does affect the attacks of percussive sounds in a way that is hard to undo. I would rather use a 2 to 5 msec delayline and live with the slight time delay, instead of messing about with 4 allpass filters in a row. As you say, might work on vocals, but what does it do on drums, or string sounds? You should take care that you're not building a phaser instead of a compressor.

Well, all in all, I don't find the G2 compressor modules that bad at all. Any slight improvement would cost dearly in dsp. They are not specifically designed for mastering compression, instead for the hype that 'everything must be compressed' to sound right. And then they don't do much harm.

To test your compressor you should feed it with a squarewave that is amplified by four in an amplifier module, so the square already hits the headroom when it comes into the compressor. Because a compressor might work very, very well with an input signal between +/- 64 units, but go completely haywire when the input signal amplitude exceeds e.g. +/- 128 units.

For really deep compression off-line processing, so non-realtime, is the way to go. There's lots of plugins that don't work in realtime that do a better job than any possible realtime compressor in 'hammering' every cycle in the wave file to exactly -1dB, with the purpose than 'the demo gets heard by the record company representative'. Wink
Back to top
View user's profile Send private message Send e-mail Visit poster's website
ik



Joined: Jan 30, 2005
Posts: 50
Location: North Carolina
G2 patch files: 25

PostPosted: Fri Mar 18, 2005 10:51 am    Post subject: Reply with quote  Mark this post and the followings unread

This is great staff !!
What a flexible machine! I love it.
Cool
ik.
Back to top
View user's profile Send private message
The Why Project



Joined: Dec 05, 2004
Posts: 118
Location: Ireland
G2 patch files: 3

PostPosted: Fri Mar 18, 2005 11:26 am    Post subject: Reply with quote  Mark this post and the followings unread

Hey Rob,

Thank you again for your insights Smile

In fact, my plan was to build a slightly faster AGC (think Audimax V1),
and make the Symmetra peak switchable, specific for vocals...
Another trick that I wanted to use is to drive the side-chain by a distorted
version of the signal itself, this can change (smoothen) the compression
slope dramatically.
In the end, from my experience with hardware compressors in the studio,
the compression slope is all-defining to which compressor I will use on
certain material... For instance, the green opto's tend to sound good on
male vocal, where a quirky compression slope works out pretty well most
of the time, but I often feel that female vocals are disturbed by a too
wicked compression slope... I prefer to use slow, but more predictable
VCA compression on them, or one of my old Siemens (slow) discretes.

Concerning the JoeMeek, Ted Fletcher ("Tedward the Compressor") is
a great guy and the old JoeMeek range is a great sounding, but also very
quirky range of compressors. I do know this, as I have both a SC2 Classic
compressor, and a VC1Q voice channel with a similar topology.
I actually don't think the JoeMeek opto's are that fast... in fact my
DBX 165 beats any other compressor in my studio for attack speed.
The JoeMeek does exactly what you describe in part 1 of your answer: If
I send in a 909 kick, the gain control will 'overshoot', and then start to
compress. This 'overshoot' generates a nice little (clipped?) click, which
makes your kick cut through the mix very well!

I agree that the G2's compressors aren't that bad, a lot also depends on
setup... but I have had no problems at all with setting it up within the G2
environment. Haven't really tried 'unprocessed' external sources on it
though...

Regards,

The Why Project
Back to top
View user's profile Send private message Visit poster's website
Rob



Joined: Mar 29, 2004
Posts: 578
Location: The Hague/Netherlands/EC
G2 patch files: 109

PostPosted: Fri Mar 18, 2005 11:51 am    Post subject: Reply with quote  Mark this post and the followings unread

The Why Project wrote:
Hey Rob,

Thank you again for your insights Smile

In fact, my plan was to build a slightly faster AGC (think Audimax V1),
and make the Symmetra peak switchable, specific for vocals...
<snip>
The Why Project


Ok, you seem to know very well what you're doing, good! Will wait in anticipation for the goodies you will come up with. I guess we don't need just another compressor patch, but specific compressors that are good for well defined jobs. Which seems a good job for an expert like you, someone who knows what effects he's after.

/Rob
Back to top
View user's profile Send private message Send e-mail Visit poster's website
mosc
Site Admin


Joined: Jan 31, 2003
Posts: 17550
Location: Allentown, PA
Audio files: 123
G2 patch files: 60

PostPosted: Fri Mar 18, 2005 11:55 am    Post subject: Reply with quote  Mark this post and the followings unread

This topic seems to have drifted into the realm of compressors. Here's another topic that might be of interest too: http://electro-music.com/forum/topic-5636.html
_________________
--Howard
my music and other stuff
Back to top
View user's profile Send private message Send e-mail Visit poster's website AIM Address
The Why Project



Joined: Dec 05, 2004
Posts: 118
Location: Ireland
G2 patch files: 3

PostPosted: Fri Mar 18, 2005 11:58 am    Post subject: Reply with quote  Mark this post and the followings unread

That's it Rob, one compressor doesn't fit all... even a Fairchild 660...

Re:Mosc's comment: Sorry, didn't want to hijack this thread!

Greetz,

The Why Project
Back to top
View user's profile Send private message Visit poster's website
mosc
Site Admin


Joined: Jan 31, 2003
Posts: 17550
Location: Allentown, PA
Audio files: 123
G2 patch files: 60

PostPosted: Sat Mar 19, 2005 8:09 am    Post subject: Reply with quote  Mark this post and the followings unread

Laughing Staying OT isn't the highest priority around here. I just mentioned it because I thought that other thead might be interesting to anyone interested in G2 compression. Somtimes I can split of posts from a topic to start a new topic, but it can get messy.

I think compression is quite often over used and missused, but I guess when it is used well, it isn't noticable - like women's makeup.

_________________
--Howard
my music and other stuff
Back to top
View user's profile Send private message Send e-mail Visit poster's website AIM Address
seraph
Editor
Editor


Joined: Jun 21, 2003
Posts: 12051
Location: Firenze, Italy
Audio files: 31
G2 patch files: 2

PostPosted: Sat Mar 19, 2005 8:13 am    Post subject: Reply with quote  Mark this post and the followings unread

mosc wrote:
it isn't noticable - like women's makeup.

very clever comparison, THX Mosc Very Happy

_________________
homepage - blog - forum - youtube

Quote:
Politics is the entertainment division of the military industrial complex - Frank Zappa
Back to top
View user's profile Send private message Visit poster's website
Display posts from previous:   
Post new topic   Reply to topic Moderators: Nord Modular Editors
Page 1 of 1 [15 Posts]
View unread posts
View new posts in the last week
Mark the topic unread :: View previous topic :: View next topic
 Forum index » Clavia Nord Modular » G2 Building Blocks
Jump to:  

You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
You cannot attach files in this forum
You can download files in this forum
emSynth

Please support our site. If you click through and buy from
our affiliate partners, we earn a small commission.


Forum with support of Syndicator RSS
Powered by phpBB © 2001, 2005 phpBB Group
Copyright © 2003 through 2009 by electro-music.com - Conditions Of Use