Just wondering, but do any of you use your NM strictly for processing audio, and not for actual tone generation? A friend recently mentioned his NMR mainly serves as an effects box for his guitar, so I was curious...

I use it for both... I run other synhts, guitar, melodica, samples, whatever through it... and I use it as a synth. Why limit youfself?

I've got an FX send from by board going into the Nord's external in so I can send audio thru it just by twisting a knob. The phaser and vocoder are excellent!

That being said, I mainly use mine for sounds...I have however toyed with the idea of getting a micro and just dedicating that to FX (I shoulda' never sold my micro when I got the expanded rack!).

Yes, that is exactly what I use my MM for. As a mangler for guitar it is great. I use a Boss GT-3 with the MM in the FX loop, and it is very versatile. The GT-3 is actually a really good midi controller.

Here's a couple of my latest patches:

HomeChorus is a homemade chorus module I've got it set up so I can run the audio through a volume swell into a delay patch and make pads, and then run w/o the swell to play on top of the pad

Bleep is pretty straight forward, a combo triggered filter and pitch modulation.

My only problem is that with my current setup -GT3 - MM audioInL- GT3 fx return - MM audioInR, I can't get any stereo sounds out. Have to get a 2nd MM I guess J

Nord_Modular_Conf wrote:

Apart from beeing just a synth I use my NMK as an additions to my bass plsying. It can be dist-filter, vocoder, pitch tracker whatever. Using all four slots infact a lot of processing can be made simultaneously. If you involve processing external to the NM like delays, tube dists etc as well the soundpalette offered is HUGE. To able to play both a bass and the NMK at the same time a footpedal MIDI-controller is also on my purchase list.

Both. Great for guitar. Fun with pitch-tracking and firing off other sounds, then looping these back. Oh for the ability to pass audio between slots J And MIDI Out J And... more...

Not strictly, but I do use it to process audio a lot. It's the best chorus/flanger I've ever used, great for mondo vocal processing. I also often put an external processing patch into an extra slot & feed the keyboard-type-patch through it (the 200% patch trick).

Up to now my only serious application for my MM is as an FXbox for my bass ("a Fender jazz '74", he said, while tears welled up in his eyes). I have a totally cool filter-dist patch that I found on the web for use with some electronic pick-system (I can't remember name or site), I made some minor mods to it to suit my needs. Been experimenting lately with the combination of a mic in "In-2", using its signal for control of distortion, vocoder or any other weirdness. I could try and post some from my Hotmail account and see how it works (later on, they're only in the red thingy at the moment, no backup on my PC) Furthermore, I use it in conjunction with my digital piano for 'conventional' synths purposes, just for fun & learning & probably some future home recording. And making my wife laugh at me. Sad to be so deeply misunderstood.

Speaking of bass playing, does anyone have a Bass Synth type patch?

Fun with pitch-tracking and firing off other sounds, then looping these back.

What do you mean by pitch tracking? I use a patch that uses an external signal to drive the slave input of slave oscillators, is that pitch tracking. I'd like to see what you have come up with.

I do a LOT of processing with it. Bass pedal, drum treatment, wierd DIY distortion and convolution (right word?) type things. Probably lay more tracks down through the modular as audio than directly from it purely synthesized. My usual signal chain is:

The modular is probably the most important element in sculpting the sound into the kind of mush i make. Its the only fx box i've found with the flexability to bash almost anything into the slimmest crack in a mix and still make it sound like part of the whole.

Here are some of the patches I use (live) for molesting my beautiful Fender Jazzbass tone. Most of them are not mine originally, but I've adjusted them for my use (except maybe the wrapped leslie, it might not be very different from the original). Some require a mic in in-2. Always plug the bass in in-1. Hope anyone has use for them (these are the first patches I send to the list; please don't yell if they're no good)

At risk of being off-topic, late in the discussion etc etc: some more convolution blabla.

I'm no mathematician (I'm actually a biologist), so no technical rantings. But have any of you tried some of the convolution proggies? Hog, anyone? Or Sonic Foundry's Acoustic mirror plugin? So far I've played a little with the Hog freeware proggie and it is so cool! For instance, try different cymbals as an impulse response on vocals. You get a very metallic kind of reverb result. Impulse responses can be downloaded from www.geocities.com/beamsonic, for all kinds of amps, mics and other stuff. Also, reverb-type IR's can be found on the web.

And if you look carefully, a "non-official tryout" version of the acoustic mirror plugin is downloadable from the web (I have it - guilty!!! - but have not tried it yet).

I know - this is not directly NM related, but it is really cool to experiment with.

I've used Acoustic Mirror a lot for convolution--it is very cool but you can't morph over time (or I haven't figured out how). Check out WaveWarp's tools. With these you can built your own convolution modules/effects/filters and actually apply them over time to audio files (no midi control, though. pity). You have to have a fast computer to do complicated stuff, but it's kind of like a poor person's Capybara (the Holy Grail for us poor folks). Full program sells for about $100, I think. I'm not sure of the address, but the company name is SoundsLogical.

I forgot to mention: this program is very informative: taught me a lot about signal processing in general, and how to construct DSP's

If you could get your own wavs to create impulse files, then you could crossfade 2 of these and use the result to create the impulse file. This would create a morph. I don't know too much about it as I haven't created my own impulses yet.

I think I've tried this, with very mixed results: also you still can't control the rate of morphing from one into another (with acoustic mirror that is, with WaveWarp its rather easy)

I tried the Hog program too and find it very interesting to experiment with. I used a recording of a voice as input for the Hog program and I synthesized some impulse responses on the Modular (so we are on topic again). (See Patches).

The patch "Impulse Response" uses two uncorrelated noise signals put through a stereo envelope (Courtesy David Peck). Using this as impulse wave gives a nice stereo reverb. "ImpulseResponse2" results in random reflections.

I think one of the problems of impulseresponse recordings is actually the recording itself...when you listen to the impulse files and found them not good and natural sounding at all....how a convolution can bring good results?

One of the problem is that most modern mikes are not ment to record from the distance. In former times there was certain types used to record in theaters from the ceiling... omni direktional capsules or very wide kidney capsules with a significant high boost. M55k from Neumann was one of these capsules.

I once did sampling with such a mikrophon...and what shall i say... the sample realy sounded like that what your ear got during the recording, not this distanced sound or to close sound you usually get.

The german word for theese technologie is "diffusefeld entzerrt"

Had to think about it a bit, but I can take a challenge. First of all everything that is said about convolution is very worthwhile, thankx for all who contributed to this thread.

My reasoning is as follows: To simplify the matter a bit consider the following: when multiplying every sample from an array of samples with every sample of an array that holds an impulse response and adding all the products together to get one single convoluted sample (or does one say convolved sample, or am I talking bull?) the characteristics of the impulse response are more or less superimposed over the sound characteristics of the sound sample, that is in very simplified words the idea of convolution, right? So what can be done is consider each _period_ of the waveforms in the sound sample as an impulse which is put into an transversal filter that has some fixed impulse response as a parameter table, a rude form of convolution might be possible when only convolving the separate periods. When the period of the waveform starts the impulse response is simple reset to the first value in the array. So when two delaylines are present and the filters impulse response is created on the fly by eg some hardsynced oscillators, it suffices to sync the oscillators to the start of every period in the sound sample. Of course it doesn't truely hold as with this idea the transversal filters' delaylines lengths should probably have to be readjusted for every period to the periods length. Well, lets skip that and live with the lofi effect of not doing that, I don't want to be theoretically ferfect, I want to create some soundeffect from the thing. So, the thing to do is to create two delaylines where every sample in those lines is available for a multiplication and the final addition. There is some trick to do just that.

Next consideration is on which part of the spectrum the effect should work. The most interesting part is definitely between 500 Hz and 3500 Hz as most of the interesting formants to mangle will most probably be in that band. Which means the filter could be limited to this band. This makes a samplerate of around 12kHz possible. Now it gets more serious as with two S&H cascades, two delaylines with the multipliers become in sight. Still the lines will be very short, but as the sample rate is reduced some effect might still be expected.

In the implementation two 12 dB filters are first used to split the audio range in three bands, the middle band is the one to be processed. The S&H delaylines use a technique called double clocking as this is needed to create a reliable cascade at the blue sample rate. The S&H actually completely operates on 96kHz, (also its yellow input! yes, that's true and confirmed by the only guy that knows, the programmer of the module). I won't go into the details why straight S&H cascades still don't work, but trust me they don't, double clocking is the only way to go. In the patch the S&H outputs of the one array are multiplied by their corresponding S&H's in the other array and their products are mixed to get a running average.

Making cascades of eg non-inverting inverter modules and multipliers could probably do a much better job, but as those all work with the 96kHz clock one needs six times more modules to get the same filter response.

I like that idea of filtering the single waveform periods each with an impulse response. By separating the periods and expand their length to the length of the impulse response table by filling the end with zeros, doing the computations period by period and chain the periods together again mixing the extra tail created by the adjustment to the impulse response tablelength in an overlapping way over the beginning of the next processed period, it might be possible to do some efficient real time transversal filtering. Simply as every period is an impulse, right? An impulse doesn't have to be just a single sample, everything that kicks is an impulse. So the data reduction is that now a convolution only has to be done for every waveforms period instead of every sample from the sample array. Or am I totally wrong here? I don't think I ever read or heard about this idea, if anyone did or can explain why this could not work, I would be pleased.

Also a straight red version that multiplies and accumulates the last 32 samples. Note the reversed order of the switches to get the computational order right. Another one where a resonator is added to the right input. And the idea of the FM_Resonator, which can do some nice effects, also with envelope follower instead of the LFO.

My thoughts about convolution!

If you think about the input as a block of shifted and scaled impulses (ie each sample is an impulse), then you can think of the output as shifted and scaled impulse responses added togeather (for each shifted and scaled input impulse you get and shifted and scaled impulse response on the output).

So to make a 10 sample delay you simply have an impulse response where the first 9 samples is zero and the 10th is one.

To make a lowpass filter, let the impulse response be of the shape of an parabola or something similar, and the sum of the samples in the impulse should be one. Think about it, for each input sample you average it with the nearby input samples, by some waighting (the parabola impulse response). Note that the signal will be a bit delayed.

To make an high pass filter, have an single impulse in the impulse response, and substract the impulse response from the lowpass filter, ie take the orginal signal and substract the lowpassed signal, ie substract the low frequencies from the original signal.

If you pop a ballon in a church and record the reverb, that pop is an aproximation of a single impulse, and the reveb is the impulse response!

When you convolve a signal with the impulse response you recorded, you are applying a scaled version of that reverb to each sample in signal (scaled by the amplitude of the impulse in the signal), so convolution is a way to work with one sample at a time in the signal!

Also a FIR filter is just a convoluiton!

Hmmm, hope you can understand what I am saying.

Had to think about it a bit, but I can take a challenge.

Wow!! You mean you actually had to think a bit and it didn't just come oozing out of you in the easy and confident "No-Rob- I'm-not-in-the-LEAST-bit-jealous" way that patches and information usually come pouring out of you?????!!!! Congratulations to the challenge-setter is all I can say. Good job. Please feel free to come up with another challenge so that Rob can send us all another patch very soon.

My reasoning is as follows: To simplify the matter a bit consider the following: when multiplying every sample from an array of samples with every sample of an array that holds an impulse response and adding all the products together to get one single convoluted sample (or does one say convolved sample, or am I talking bull?)...

You are not talking bull and it would be a convolved sample. A convoluted sample is something slightly different. I.e a convolved sample may also be convoluted, but not all convoluted samples are convolved. (If you really want to know what convoluted is, you should have a pretty good idea once you have worked your way through the previous sentence!)

... the characteristics of the impulse response are more or less superimposed over the sound characteristics of the sound sample, that is in very simplified words the idea of convolution, right?

As far as I know, right. Multiplication is just a long-winded way of adding, etc. etc .... so in the end integration is "just" addition. Ho, ho, ho!!

So what can be done is consider each _period_ of the waveforms in the sound sample as an impulse which is put into an transversal filter that has some fixed impulse response as a parameter table, a rude form of convolution might be possible when only convolving the separate periods.

True, given the relationships between the time and frequency domains that I asked about earlier. (Notice the extremely cunning way in which I am trying to show not only what an extremely intelligent fellow I am, also ... but what an essential and vital contribution I made to your patch!! Weren't fooled for a minute, though, were you!!!)

OK. Cunning.

So when two delaylines are present and the filters impulse response is created on the fly by eg some hardsynced oscillators, it suffices to sync the oscillators to the start of every period in the sound sample.

Got it. But ...

Of course it doesn't truely hold as with this idea the transversal filters' delaylines lengths should probably have to be readjusted for every period to the periods length.

Exactly.

Well, lets skip that and live with the lofi effect of not doing that, I don't want to be theoretically ferfect, I want to create some soundeffect from the thing.

Good approach. A good mathematician always begins to establish a model by stripping things down to their bare and most essential parts. Once those are correct, details and substructures can gradually be added.

So, the thing to do is to create two delaylines where every sample in those lines is available for a multiplication and the final addition. There is some trick to do just that.

Nice transfer of the modus operandi. Instead of manipulating the frequencies, we mess about with delays i.e. the time domain, which is considerably more easily dealt with in the NM's architecture. Even _I_ can understand that once some other smart fellow has pointed it out.

Next consideration is on which part of the spectrum the effect should work. The most interesting part is definitely between 500 Hz and 3500 Hz as most of the interesting formants to mangle will most probably be in that band.

Well ... being a man who likes the play between rhythms, I would have liked something a little lower, but we can't have everything.

Which means the filter could be limited to this band. This makes a samplerate of around 12kHz possible. Now it gets more serious as with two S&H cascades, two delaylines with the multipliers become in sight. Still the lines will be very short, but as the sample rate is reduced some effect might still be expected.

Agreed.

OK. Obviously, though, those of us who would like different aspects and audio streams to play with can adjust this to suit our tastes. I shall do so when I get back home again to my machine.

The S&H delaylines use a technique called double clocking as this is needed to create a reliable cascade at the blue sample rate.

... i.e. data transfers on both the rising and the falling edges of the clock? This is because data losses and interference increases with frequency, correct? Therefore, double clocking makes the timing tighter?

The S&H actually completely operates on 96kHz, (also its yellow input! yes, that's true and confirmed by the only guy that knows, the programmer of the module). I won't go into the details why straight S&H cascades still don't work, but trust me...

Now ... Why should I trust you???!!!!!!!

... they don't, double clocking is the only way to go.

OK....I'll trust you... But only on this. I'm still not trusting you with the details to my bank account, OK!!

In the patch the S&H outputs of the one array are multiplied by their corresponding S&H's in the other array and their products are mixed to get a running average.

Yes ... that would seem to be the basic idea. Good solution ...

Making cascades of eg non-inverting inverter modules and multipliers could probably do a much better job, but as those all work with the 96kHz clock one needs six times more modules to get the same filter response.

... and also efficient.

Anyway, if this patch is entitled to be called real convolution, I don't know.

I think that we have already settled that convolution is a very poetic word that people are entitled to use as they wish as long as they are precise and correct in its usage.

Much blood was spilt on this point.

You must have been asleep at the time!!

Looks good, anyway. That's always a good start. Can't tell you what it sounds like right now, though.

I like that idea of filtering the single waveform periods each with an impulse response.

Yes. It's a nice idea.

By separating the periods and expand their length to the length of the impulse response table by filling the end with zeros, doing the computations period by period and chain the periods together again mixing the extra tail created by the adjustment to the impulse response tablelength in an overlapping way over the beginning of the next processed period, it might be possible to do some efficient real time transversal filtering.

Phew. Shove the end upon the beginning, and you have a lot of intersecting universes working out their karma because each succeeding universe is affected by the unfinished business left over from the previous one. That about right?

Simply as every period is an impulse, right?

That's my view, also.

An impulse doesn't have to be just a single sample, everything that kicks is an impulse.

Agreed. Don't know what the purist mathematics buffs would make of it, though. Wait a minute ... you're one of them, aren't you?!!!

So the data reduction is that now a convolution only has to be done for every waveforms period instead of every sample from the sample array.

Very clever, Rob. Nice.

Or am I totally wrong here?

Well I think it's more that it's kind of a poor man's convolution overlaid on the non-convoluted originals. It's probably a convolution in the same kind of way that differencing is differentiation.

I don't think I ever read or heard about this idea, if anyone did or can explain why this could not work, I would be pleased.

I'm sorry that I can't really please you because the principles seem fine to me.

But if you want to be made happy, then here's what I would say ...

Looks to me that what you have ACTUALLY done is to directly sum an integral rather than integrate and so therefore this is not a convolution. Did that make you jump for joy? Glad to have made you happy, Rob. Hope I can do so some time very soon again!! You have also restricted the frequencies over which the procedure has been conducted by restricting yourself to certain time intervals, but there,s nothing intrinsically wrong with that. This is a perfectly acceptable procedure if you ask me ... although I suppose that TECHNICALLY it isn't a convolution because TECHNICALLY you haven't actually integrated. You have simply summed and then taken differences which is not TECHNICALLY the same thing ... but I for one don't care in the slightest. This is in my view a nice and elegant solution which realizes the principles involved in a nice practical way ... and I would indeed call it a convolution, but then again I'm USUALLY (but not always) more one of those poetic types.

Exactly what did your Mummy feed you when you were younger, by the way??!!!

I had fun figuring this out.

That's nice to hear!!! I had fun trying to work out how it worked and wondering what it would sound like.

Don't suppose they're putting any idiot lights on the modules in Version 4, are they? You know ... so that the modules light up in Christmas Tree fashion as they get used and flashing lights switch on and off so that total ignoramuses like me can get an idea of what's happening? I could do with some of those to work my way through some of the patches that you and Jan Punter and Dave Peck and people like that send in here sometimes!!

... i.e. data transfers on both the rising and the falling edges of the clock? This is because data losses and interference increases with frequency, correct? Therefore, double clocking makes the timing tighter?

This has to do with the calculation order for the modules, one would think that if S&H modules are simply cascaded and clocked by the same pulse every value in the array would shift one position, but in fact the input would be shifted straigth to the last S&H. When placing the S&H in reversed order it still doesn't work, which has to do with the mixed internal sample rate of red and yellow/blue signals and how they interrupt each other. So, double clocking is like a true bucket brigade, where the even numbered persons have to turn backwards to receive a bucket from the odd numbered persons that have to turn forward to be able to handle over their bucket, after wich the odd numbered ones have to backwards and the even numebered ones forward with their just received bucket, etc. This reduces the number of buckets travelling along the brigade to half the number of people in the brigade. Having the same number of people as the amount of buckets and all people turning forward trying to handle over their bucket to the next person would cause lots of wet trousers but the building would most probably burn down completely.

I'm still not trusting you with the details to my bank account, OK!!

Phew. Shove the end upon the beginning, and you have a lot of intersecting universes working out their karma because each succeeding universe is affected by the unfinished business left over from the previous one. That about right?

One thing definitely leads to another, the only way to bend that chain seems to be a change in attitude, well, this is philosophical and debateable, I know. But it suggests an interesting possibility of creating iterating processes that can be steered by changing the impulse response that drives these processes for only a single iteration and then reuse the original impulse resonse again. Something like the 'one man can make the difference' idea. I suppose that is common in drumming, where if one beat is changed all drummers might get inspired to a new rhythm, or to quickly search for the bar.

Agreed. Don't know what the purist mathematics buffs would make of it, though. Wait a minute ... you're one of them, aren't you?!!!

Actually no! The only thing I'm interested in is: can all these numbers be turned into music and art. I know that mathematicians can experience the aesthetics of math, the old greeks already did (and so did many other old cultures), so is it interesting to try to make that aesthetic more accessible to the general public. (In my language 'esthetiek' is the principle of applied aesthetics, so perhaps 'aesthetic' exists as a singular word in English too?) Just for a change between the burger kings, six packs and all that other stuff we're made to believe that makes the world go round.

Anyway, in the virtual digital world an impulse might be considered a pulse approaching zero width in time, infinite energy and a surface of one, but in the real analog world the impulse response to someone kicking my butt is 'auw'. So, real impulses have character and at least several samples are needed to capture that character. That's why the complete sampled impulse response can be used as the impulse in physical modelling of sounds to excite an otherwise characterless resonator and create a sound with a definite character.

... and I would indeed call it a convolution, but then again I'm USUALLY (but not always) more one of those poetic types.

Don't believe the scientists, poetry is more essential...

Exactly what did your Mummy feed you when you were younger, by the way??!!!

Amongst other totally disgusting things a spoonful of whale fishoil every day. Which is now forbidden by law as it turned out feeding children fishoil strongly increases the chances of them becoming revolting revolutionairs, or even worse: long haired artists.

I had fun trying to work out how it worked and wondering what it would sound like.

So far, we've got things like physical modeling, midi-sync'd LFOs, lo-fi sampling, 200% patches, soft sync, arpeggiators, non-western scale systems, etc. Nice!

Rob Hordijk wrote:

This has to do with the calculation order for the modules, one would think that if S&H modules are simply cascaded and clocked by the same pulse every value in the array would shift one position, but in fact the input would be shifted straigth to the last S&H.

Man but do I feel stupid. I once tried to make an emulation of an African vocal ensemble using an equiheptatonic scale and I just couldn't get it to work. I should have thought to send a question in, but I just abandoned it for another project I was working on. Half the things I try don't work anyway, but usually I just blame myself. Nice to know that sometimes it's the tools and not the workperson.

When placing the S&H in reversed order it still doesn't work, ...

I discovered that the hard way!! ...

.. which has to do with the mixed internal sample rate of red and yellow/blue signals and how they interrupt each other.

... but I never got as far as thinking it out in this way.

So, double clocking is like a true bucket brigade, where the even numbered persons have to turn backwards to receive a bucket from the odd numbered persons that have to turn forward to be able to handle over their bucket, after wich the odd numbered ones have to backwards and the even numebered ones forward with their just received bucket, etc. Now even _I_ can understand that one. Thank you for the explanation. (Gives me an idea for a 'bucket brigade' patch!! I'll try it when I get back).

This reduces the number of buckets travelling along the brigade to half the number of people in the brigade.

Of course.

Having the same number of people as the amount of buckets and all people turning forward trying to handle over their bucket to the next person would cause lots of wet trousers but the building would most probably burn down completely.

Yup. My trousers got VERY wet when I tried this. (!!)

My wife just thought that I was getting -- well --- a little over-excited and she hinted that maybe I would get even more excited if I would turn my modular off instead and come to bed!!!!

Yup. Those days have already arrived. My older daughter is nearly 17.

Rob wrote:

By separating the periods and expand their length to the length of the impulse response table by filling the end with zeros, doing the computations period by period and chain the periods together again mixing the extra tail created by the adjustment to the impulse response tablelength in an overlapping way over the beginning of the next processed period, it might be possible to do some efficient real time transversal filtering.

then Friday's Child (that's me!) wrote:

Phew. Shove the end upon the beginning, and you have a lot of intersecting universes working out their karma because each succeeding universe is affected by the unfinished business left over from the previous one. That about right?

then Rob wrote:

One thing definitely leads to another, the only way to bend that chain seems to be a change in attitude, well, this is philosophical and debateable, I know.

Yeah! Maybe we should establish the Nord Modular OT/Religious/Philosophy List for that one!!

But it suggests an interesting possibility of creating iterating processes that can be steered by changing the impulse response that drives these processes for only a single iteration and then reuse the original impulse resonse again.

You know something Rob ... I actually understood that. Oh yes I did! I know that you deliberately speak that way in order to make me feel stupid, but that time it didn't work. I followed every word!! You must be losing your touch.

Something like the 'one man can make the difference' idea.

You know something ... I understood that too. You stole it from George Bush, didn't you??!!

I suppose that is common in drumming, where if one beat is changed all drummers might get inspired to a new rhythm, or to quickly search for the bar.

Yes. That is pretty much what happens. One beat being changed can easily be like a modulation to a completely new key which can make a whole host of new notes available for the musicians. This is why, on more formal occasions, a master drummer is appointed to lead the proceedings. It's not necessarily that he or she is any better than the others. More a way of making sure that everyone changes chords in an agreed way.

Don't know what the purist mathematics buffs would make of it, though. Wait a minute ... you're one of them, aren't you?!!!

Actually no!

(The erstwhile Nord Patch Programmer sometimes known as Friday's Child coughs discretely and unbelievingly into his sleeve.)

The only thing I'm interested in is: can all these numbers be turned into music and art. A long time ago, they were all the same thing!

I know that mathematicians can experience the aesthetics of math, ...

Yes. Unfortunately, my own skills in mathematics are deplorable ... but I still like the aesthetics of it. We are all weird somehow, I think!

(In my language 'esthetiek' is the principle of applied aesthetics, so perhaps 'aesthetic' exists as a singular word in English too?)

It does exist as a singular word in English, but it doesn't have quite the same effect that way in English as it seems to have in Dutch. That's what I think, anyway. Quite often when it is used in the singular, it is an adjective. It can also be used as a singular noun in which case, as far as I can tell, it would then refer to a particular school or a particular style or taste or "aesthetic", rather than distinguishing between the theory and the practice of aesthetics. But with the argumentative nature of the people on this list, I am quite sure that someone here will speak up very soon and disagree with me totally!! That then means that you are going to have to choose who to believe. If I were you, I'd know who my choice would be.

I don't believe all that advertising for a moment. Back in my home village there's this old man stirring a big pot of soup. If you ask him why he's doing it he tells you that he can't stop stirring right now because if he did then the world would stop turning. He is still stirring that huge pot of soup ... and the world is still spinning round and round. I therefore rest my case. I believe that mathematicians would call that a constructive proof!!

Music seems to be a good translator.

Agreed.

Well, for ages there has been quite a tight relationship between music, math, the position of the stars and faith, until modern western science deleted music, star positions and faith from this list.

Also agreed. I don't know why they allowed it to happen, but musicians have allowed themselves to be marginalised away from many of those grand explanations. This would not be so bad if music had not allowed itself to get bogged down in e.g. singing endless songs about people's love lives (the only exceptions surely being songs like "I lurv ma modular babeee (noodle, noodle noodle); it's so cool and so red (noodle, noodle, noodle, NOOO-OOO-dle). It would be nice for a good musician e.g. to be able to almost unseat a government by writing e.g. an anti-war song or some such. People do write such songs I know ... but on the whole the audiences don't seem to want to listen to them.

Still. Mustn't get TOO OT.

These deletions tend to reduce the human individual to a mechanical, economical production unit without any purpose of being at all. I don't think those deletions are entirely responsible, but that's a whole other -- and very interesting -- story.

And if this would truely be the case then I would suggest to kill this planet right away instead of having to slowly suffocate in our pollution for some more decades, as this suffocation indicates a quite bizarre form of masochism. And me, I'm not at all into masochism!

Maybe ... if you got into masochism you would enjoy things more!!!

As above, have you tried masochism? It's getting a very good press.

Anyway, in the virtual digital world ...

Aaah that!!

... an impulse might be considered a pulse approaching zero width in time, infinite energy and a surface of one,

... in other words ... a spike, no? A random discontinuity in the field of sound.

Just as an aside, isn't this also the definition of a Dirac delta function, and the way in which Paul Dirac solved the then outstanding problems in quantum physics?

... but in the real analog world the impulse response to someone kicking my butt is 'auw'.

LOL

So, real impulses have character and at least several samples are needed to capture that character.

Yes ... they are continuous and must extend over a 'perceptible' period of time to be of interest to us humans. However, the quantum domain is also "digital", I believe, and seems to have no trouble being the foundation of everything. Well ... that's what the scientists would have us believe, but methinks they haven't drunk enough soup!!