(computer) Music at its most basic.

It's only cool if no one's heard of it.

Moderators: SecondTalon, Moderators General, Prelates

User avatar
Midnight
Posts: 2170
Joined: Mon Dec 10, 2007 3:53 am UTC
Location: Twixt hither and thither. Ergo, Jupiter.

(computer) Music at its most basic.

Postby Midnight » Mon Oct 11, 2010 10:18 pm UTC

So I dunno, I was fooling around with sine waves in audacity, and am currently taking a class on music production via computers--it's fairly basic and covers sampling, sequencing, loops, patches, MIDI, Reason, Pro Tools, and mixing.
It does not, however, cover the foundations of synthesizing--that is to say, if I want to learn how to make a C major chord (C E G C) using just audacity-generated sine waves, I'm SoL.
I know Dream has some top tier degree in regards to THIS VERY THING, and I know there's plenty of people that possess more expertise than I, so I thought "Hey, why can't we do this together!" I'm sure I can go anywhere on the internet and find a guide to manipulating overtones into a desired chord or sound, but I like you all, so I thought that this thread could serve as a database, or better yet a forum, for discussing the creation of sounds using nothing but computers. The techno thread doesn't cover this, the music sharing thread (of your own or otherwise) doesn't cover this, the various music theory ones don't cover this... so this one will.

I'll start:
I know enough of math (very little) and music (slightly more) that a sine wave at 440 will make an A, and one at 660 will make an E; a perfect fifth. However, the only way I can craft a perfect fifth using just audacity (after ~45 seconds of experimentation) is by putting the E panned far left and the A far right (or vice versa). If they come out of the same channel, I get a rather harsh overtone. According to wikipedia, the stacking of sine waves gets me something like an A and all of it's octaves--220, 440, 880, and so forth.
That's acceptable, BUT if I add in, say, the perfect fourth of A, D (mathematics telling me it's 586.667hz) I get a dull buzz.
I assume a low-pass filter will get rid of the harsh overtones with the perfect-fifth dealio, but I rather doubt it will eliminate my buzz.
SO, if I want to make a chord, first-third-fifth, how do I do that without nasty buzzes? Is there a way? Must I use weird frequencies to manipulate the overtones to my advantage? Would some combination of x-pass filters work? It doesn't matter if it's an A chord; if it's easier to start from a baseline of 100 or 1000 hertz, I'm down.


EDIT:
Two more minutes of experimentation reveal that if I just drop the gain a few decibels, to -6 or so, the harshness disappears. Interesting. The more sine waves I add, though, the more the harshness comes back, so I need to drop the gain even further. Which could be a vicious cycle, but I'm not really planning on polychords, here.
EDIT TWO:
http://www.phy.mtu.edu/~suits/notefreqs.html this is helpful-ish. Pretty sure the decimal points aren't quite as accurate as I need them to be. Switching from manipulating an E @330 (Ab at 412.5, B at 495) to those new numbers introduces a bit of a wobble. I'll stick with numeric perfection.
uhhhh fuck.

Роберт
Posts: 4285
Joined: Wed May 14, 2008 1:56 am UTC

Re: (computer) Music at its most basic.

Postby Роберт » Tue Oct 12, 2010 4:02 pm UTC

Midnight wrote:I'll stick with numeric perfection.
Wait, are you wanting to use just intonation instead of the more typical* equal temperament?

*in modern western music

Midnight wrote:Two more minutes of experimentation reveal that if I just drop the gain a few decibels, to -6 or so, the harshness disappears. Interesting. The more sine waves I add, though, the more the harshness comes back, so I need to drop the gain even further. Which could be a vicious cycle, but I'm not really planning on polychords, here.

I don't mean to sound rude by asking a basic question, but it is possible you're clipping?
The Great Hippo wrote:[T]he way we treat suspected terrorists genuinely terrifies me.

User avatar
Dream
WINNING
Posts: 4338
Joined: Tue Dec 04, 2007 7:20 pm UTC
Location: The Hollow Scene Epic

Re: (computer) Music at its most basic.

Postby Dream » Tue Oct 12, 2010 4:14 pm UTC

It's great to hear someone's interested in these things. Computers do far more in music than just editing, auto-tuning, and letting talentless DJs beat match.

Doing the things you're describing is working like Stockhausen and the Elektronische school in Germany. It's fascinating, and capable of producing some exceptional music. But I wouldn't work that way today (personal here, many people still work exactly this way) because it doesn't leverage the power of computers to do the arranging and mixing. It works the same way tape studios did. Instead, I'd dive straight in and learn either Supercollider (if you like CLI programming) or pureData (if you like graphical object oriented environments). These are programs of limitless potential, but require a little programming nous to exploit fully. Still, if you're generating sine waves experimentally, they'll do the trick far better than a multitrack editor will, and the skills you gain with them will stand to you as long as you're computer musicing. This is a track that uses a Supercollider designed sine wave generator to layer dozens of waves over one another within different frequency and density parameters. There's a breakdown at 1.30 that highlights the sound. It's a few lines of code, but would take 60 to 130 tracks in an editor, and hours of transposition and gain setting to create a single sound. The Elektronische guys would have done that, but I prefer to let the computer take care of it.

The great thing about pD, or one of the many great things about it, is that the author of the program, Miller Puckette, is also the author of this standard text on computer music, in which all the examples are created in pD. If you can handle the dry technical language, this is an invaluable resource.

Midnight wrote:Two more minutes of experimentation reveal that if I just drop the gain a few decibels, to -6 or so, the harshness disappears. Interesting. The more sine waves I add, though, the more the harshness comes back, so I need to drop the gain even further. Which could be a vicious cycle, but I'm not really planning on polychords, here.

This is a digital signal issue, rather than an acoustic one. You have a finite number of bits in each audio sample, either 24 or 16 for most systems. If you go outside that finite range, for instance by summing two signals, each of which uses more than half the available bits, you get distortion. This distortion is the harshness you're hearing. The reason it goes away when you hard pan the signals is that each channel has its own 16 or 24 bits, and assigning one signal each to left and right means there is once more a full complement of bits available to work with per signal.
http://www.phy.mtu.edu/~suits/notefreqs.html this is helpful-ish. Pretty sure the decimal points aren't quite as accurate as I need them to be. Switching from manipulating an E @330 (Ab at 412.5, B at 495) to those new numbers introduces a bit of a wobble. I'll stick with numeric perfection.

And this is an acoustic problem. Temperament, the precise tunings that make up a scale of notes, is not a fixed concept. So a piano Ab may not be the same tuning as the mathematically derived interval between the root pitch (usually A at 440hz) and itself. A concert Ab may well be closer, made as it is with wind and stringed instruments that have essentially infinite pitch granularity. Working with computer systems, you will have the choice, and will always have to make the choice, between perfect and compromised tuning. Both have their strengths and weaknesses, which are well documented elsewhere. Your wobble may be "beating" which is the phenomenon of periodic shifts in amplitude caused by phase relationships between to similar-shaped waves.
Midnight wrote:Must I use weird frequencies to manipulate the overtones to my advantage?

No, but it helps. Creation of overtones is in essence the manipulation of what we perceive as timbre. Overtones are nothing more than the product of interference between sine waves. So as you try to make interesting sounds by synthesis of various kinds, the interaction of waves generating upper partials becomes very important.
I knew a woman once, but she died soon after.

Роберт
Posts: 4285
Joined: Wed May 14, 2008 1:56 am UTC

Re: (computer) Music at its most basic.

Postby Роберт » Tue Oct 12, 2010 4:36 pm UTC

Wow, Dream, that's some cool stuff. I like it.

@Midnight: You found a solution to your clipping problem. Turning it down. If you need to turn it down by 6dB to fit a second sine wave (-3dB is half power, so -6dB is half voltage, that sounds right in my head), than
to fit 4 sine waves, you'd need the waves to be at -12,
to fit 8, you'd need them to be at -18,
to fit 16, you'd need to be at -24.

You could also use a soft limiter to remove the artifacts (it would be more like saturated tape distortion instead).

From http://www.sweetwater.com/expert-center ... oftLimiter
Soft limiting is basically applying a soft knee type of process to a limiter. Soft limiting rounds the peaks of audio to allow a hotter signal to be printed to tape or disk without clipping. It comes in especially handy for digital recording. Digital converters are more accurate at higher levels (up to clipping) so applying a soft limiter to the signal allows the overall signal to be captured several dB hotter while not sounding overly compressed (only the peaks are altered). In many ways a properly configured soft limiter will emulate the natural characteristics of analog tape saturation and can thus produce a very pleasing and familiar sound in digital recording.

There are some free vsts that do it, so if want to experiment, you might try that, to.

I think some synthesizer theory might be helpful if you don't know much. Try http://www.audiogeekzine.com/2010/07/sy ... -lesson-1/

Keep in mind all waves, square, sawtooth, etc, are essentially a collection of sine waves - one at the root frequency, and an infinite number at harmonic frequencies. Distortion on a guitar is generally similar, as well - the signal isn't linearly amplified, so the root frequency gets some harmonics added in.
The Great Hippo wrote:[T]he way we treat suspected terrorists genuinely terrifies me.

User avatar
Dream
WINNING
Posts: 4338
Joined: Tue Dec 04, 2007 7:20 pm UTC
Location: The Hollow Scene Epic

Re: (computer) Music at its most basic.

Postby Dream » Tue Oct 12, 2010 5:11 pm UTC

Роберт wrote:to fit 4 sine waves, you'd need the waves to be at -12,
to fit 8, you'd need them to be at -18,
to fit 16, you'd need to be at -24.
I see how your thinking works, but it's not quite that simple. Here's a Logic session with nine sine waves at 1000hz, 0dB, requiring 6.4db of attenuation on the aux bus to give -0.1db output.
Picture 11.png

If I understand your deduction, you've assumed a linear scal of intensity, where dB is logarithmic. You seem to be correct in terms of relative sound pressure level, but those levels are represented with different dB values than you've used.
Роберт wrote:You could also use a soft limiter to remove the artifacts (it would be more like saturated tape distortion instead).

This would work perfectly, but unless the signal is dynamic, as in its intensity changes over time, the limiter won't do much that a volume control wouldn't. Limiting will lower the level below the clipping point, but so will turning down the master fader. On the other hand, if you have a dynamic signal, the limiter will turn it down only when it would otherwise have clipped, leaving it alone in quieter moments, which is very useful. A constant level sine will of course always have the same intensity, so limiting is basically just attenuation.
Роберт wrote:Wow, Dream, that's some cool stuff. I like it.

Thanks! That's from my old studio based production days, so I don't really make music like that anymore, but it's nice to hear a compliment. :)
I knew a woman once, but she died soon after.

Роберт
Posts: 4285
Joined: Wed May 14, 2008 1:56 am UTC

Re: (computer) Music at its most basic.

Postby Роберт » Tue Oct 12, 2010 7:30 pm UTC

Dream wrote:If I understand your deduction, you've assumed a linear scal of intensity, where dB is logarithmic. You seem to be correct in terms of relative sound pressure level, but those levels are represented with different dB values than you've used.
Actually, I think I see what I did wrong. If you had two sounds that were 0dB, and put them together, the output will be at 6dB if they are the exact same signal, -Inf dB (silence) if they are the same signal except the phase is inverted for one of them, and potentially anything in-between, depending how the signals interact. Right?

Dream wrote:This would work perfectly, but unless the signal is dynamic, as in its intensity changes over time, the limiter won't do much that a volume control wouldn't. Limiting will lower the level below the clipping point, but so will turning down the master fader. On the other hand, if you have a dynamic signal, the limiter will turn it down only when it would otherwise have clipped, leaving it alone in quieter moments, which is very useful. A constant level sine will of course always have the same intensity, so limiting is basically just attenuation.

Well, the limiter would in that case basically be like having automatic volume automation, so it would be useful for when playing around in that you wouldn't have to keep adjusting the volume. But I don't think that's right. A limiter I thought softy clipped the waveform. That is, if I was playing a notes at 220Hz, 330Hz, 440Hz, 660Hz and 880Hz, the highest peak in my waveform will be occurring when all the sine waves line up, which would be around 73 times a second. I thought a soft limiter would apply soft clipping to those peaks, allowing the signal to be overall louder and producing less harsh distortion at those peaks. This would be similar to the hard clipping he's getting already except the harmonics produced from the distortion would be more pleasing.
The Great Hippo wrote:[T]he way we treat suspected terrorists genuinely terrifies me.

User avatar
Midnight
Posts: 2170
Joined: Mon Dec 10, 2007 3:53 am UTC
Location: Twixt hither and thither. Ergo, Jupiter.

Re: (computer) Music at its most basic.

Postby Midnight » Wed Oct 13, 2010 2:08 am UTC

Роберт wrote:
Midnight wrote:I'll stick with numeric perfection.
Wait, are you wanting to use just intonation instead of the more typical* equal temperament?

Just intonation. If I'm going for an E, I go 440hz (A, my baseline) x 1.5 (the ratio for a perfect fifth) = 660hz, an E above A. Not 656.8998181 or whatever. In the absence of a keyboard, it doesn't really matter. A difference of a few cents isn't noticeable to a human ear, and the only instruments I'd use alongside these sine waves are guitars/basses, which can be tuned to match. Slightly, at least.

Роберт wrote:
Midnight wrote:Two more minutes of experimentation reveal that if I just drop the gain a few decibels, to -6 or so, the harshness disappears. Interesting. The more sine waves I add, though, the more the harshness comes back, so I need to drop the gain even further. Which could be a vicious cycle, but I'm not really planning on polychords, here.

I don't mean to sound rude by asking a basic question, but it is possible you're clipping?

I'm guessing Dream's right, but regardless of whether the answer is his response or clipping, turning it down solves the problem so I dun give a fuck. I'm not clipping though; the tones generated in Audacity line up perfectly to 0db or whatever.


I might download a soft limiter plugin, but I'm liking the idea of just using the most baseline equipment possible to make Weird Shit. Such as manipulating Audacity's hilariously awful reverb and echo to get some old-fashioned sonic chaos.
uhhhh fuck.

Роберт
Posts: 4285
Joined: Wed May 14, 2008 1:56 am UTC

Re: (computer) Music at its most basic.

Postby Роберт » Wed Oct 13, 2010 3:37 am UTC

Midnight wrote:I'm guessing Dream's right, but regardless of whether the answer is his response or clipping, turning it down solves the problem so I dun give a fuck. I'm not clipping though; the tones generated in Audacity line up perfectly to 0db or whatever.

Dream was kind enough to explain what clipping was instead of just saying "you're clipping", but he and I are saying the same thing here. Your individual waveforms are 0dB, but when the are summed the final waveform goes beyond the range, thus producing the unexpected distortion. http://en.wikipedia.org/wiki/Clipping_(audio)#Digital_clipping

However, since you're just wanting to have fun, don't worry about the theory - just experiment and see what kind of crazy sounds you can make. Your filter idea could be fun, particularly if you put some automation on the filter to move it around during the sonic assault.
The Great Hippo wrote:[T]he way we treat suspected terrorists genuinely terrifies me.

User avatar
Dream
WINNING
Posts: 4338
Joined: Tue Dec 04, 2007 7:20 pm UTC
Location: The Hollow Scene Epic

Re: (computer) Music at its most basic.

Postby Dream » Wed Oct 13, 2010 1:25 pm UTC

Роберт wrote:That is, if I was playing a notes at 220Hz, 330Hz, 440Hz, 660Hz and 880Hz, the highest peak in my waveform will be occurring when all the sine waves line up, which would be around 73 times a second. I thought a soft limiter would apply soft clipping to those peaks, allowing the signal to be overall louder and producing less harsh distortion at those peaks.

That's how a limiter works in theory, but they don't work that fast. They work on transients, the dynamic peaks in the overall sound, rather than the peaks in the waveform. 73 events per second is well into the audio range, and would be interpreted by the limiter's programming as a constant, rather than a dynamic sound.As a rule of thumb, events discernable to the ear as being separate can be limited, and those that can't are interpreted as a single tone.

Also, a limiter doesn't clip the waveform in a pleasing manner. It ducks the signal intensity to prevent clipping altogether. A clipped sine wave looks like this:
Picture 12.png
While the limited looks like this:
Picture 13.png
The limited wave would be identical to the original undistorted wave, not a less drastic version of the first. By lowering the intensity of the wave to a level that can be handled by the software, the clipping is eliminated altogether.

Midnight wrote:A difference of a few cents isn't noticeable to a human ear, and the only instruments I'd use alongside these sine waves are guitars/basses, which can be tuned to match. Slightly, at least.

The difference may not be noticeable, but the interference patterns it generates will be. So while you can quite happily ignore the difference between 440 and 440.01, you may have to stop ignoring it when you begin to stack many waves that will give rise to unexpected frequency relationships between their overtones. Creatively you're on solid ground, just do what sounds good. Technically, it's probably worth giving yourself a mechanism to retune as necessary to eliminate unwanted beating.
Midnight wrote:I'm not clipping though; the tones generated in Audacity line up perfectly to 0db or whatever.

Once the signal has been clipped, the level can't exceed 0dB, by definition. This is because clipping (for most purposes) only happens on the way out of the system. The internal architecture of most audio software is not the 16 or 24 bit fixed point or CDs. They use a 32 bit floating point architecture. This means that they can have essentially limitless gain applied within the system, as long as that gain is removed before the conversion to 16 or 24 bit for output. As in this example, where a sine wave starts out at 0dB (oscillating between -1 and 1 floating point), is multiplied by 100 to clip into a square wave on output, also at 0dB and then divided by 100 to reveal not a square wave 100 times smaller, but the original sine at 0dB that was always there.
Picture 14.png

So only once your audio leaves Audacity does the distortion happen, and it is that distortion that maintains the level at 0dB. Within the system, the audio doesn't clip at all. Does that make sense?
I knew a woman once, but she died soon after.

Роберт
Posts: 4285
Joined: Wed May 14, 2008 1:56 am UTC

Re: (computer) Music at its most basic.

Postby Роберт » Wed Oct 13, 2010 3:10 pm UTC

Dream wrote:That's how a limiter works in theory, but they don't work that fast. They work on transients, the dynamic peaks in the overall sound, rather than the peaks in the waveform. 73 events per second is well into the audio range, and would be interpreted by the limiter's programming as a constant, rather than a dynamic sound.As a rule of thumb, events discernable to the ear as being separate can be limited, and those that can't are interpreted as a single tone.

http://en.wikipedia.org/wiki/Limiting
It sounds like from that article, "limiting" can refer to compression type action that you are referring to, so you are right about some types of limiters behaving that way. However, I was referring to soft limiting, which would be almost trivial to implement in software* (although perhaps a challenge to a good method and implementation so that it sounds good and doesn't introduce latency or tax the processor). Soft limiting behaves similarly to a tube amplifier or natural tape compression - instead of adjusting gain with some attack an release time, it distorts the actual waveform. However, the distortion is less harsh than the distortion that occurs when you actually clip.

*psuedo code for a form of of soft limiting

Code: Select all

if absoluteValue(the32BitNumber) > 30,000
then
        if(the32BitNumber > 0)
                the16BitNumber = 30,000 + (the32BitNumber-30,000)/10
        else
                the16BitNumber = -30,000 + (the32BitNumber+30,000)/10
        endif
else
        the16bitNumber = the32BitNumber
endif

return the16BitNumber

I don't know what that would sound like, but my point is, in digital signal processing, it's easy to do limiting in that sense, rather than the compression sense.
The Great Hippo wrote:[T]he way we treat suspected terrorists genuinely terrifies me.

User avatar
Dream
WINNING
Posts: 4338
Joined: Tue Dec 04, 2007 7:20 pm UTC
Location: The Hollow Scene Epic

Re: (computer) Music at its most basic.

Postby Dream » Wed Oct 13, 2010 4:25 pm UTC

Роберт wrote:I don't know what that would sound like, but my point is, in digital signal processing, it's easy to do limiting in that sense, rather than the compression sense.

What you're describing is a synthesis technique called waveshaping. You describe what's called a transfer function, a table of data in the range -1 to 1 with values from -1 to 1, or just the mathematical function that describes the values. Then you apply your sound wave to the transfer function, and it returns the new (in your case compressed) values for each value input. It is very much not limiting, which applies the same transformation to all values, a transformation that increases its effect depending on the overall (as opposed to sample-specific) level. So limiting leaves the shape of the wave unchanged relative to itself, merely expressing it in a smaller range of values. Waveshaping alters the shape of the wave, and thus the timbre of the sound.

For the record, the transfer function that would give the effect you're after does indeed distort the output sound, but in a very useful way. As you increase the input level, the output level distorts more and more, just like in a valve system. It doesn't sound like a valve, but it behaves like one, which is good for a lot of purposes. The function looks like a shallow "S" shape, with values something like (-1, -.85) (0,0) (+1, +.85).
I knew a woman once, but she died soon after.

Роберт
Posts: 4285
Joined: Wed May 14, 2008 1:56 am UTC

Re: (computer) Music at its most basic.

Postby Роберт » Wed Oct 13, 2010 4:54 pm UTC

Why don't you fix the Wikipedia article I linked to then, since it's apparently wrong.
Soft limiting is limiting in which the transfer function of a device is a function of its instantaneous or integrated output level. The output waveform is therefore distorted, but not clipped.
The Great Hippo wrote:[T]he way we treat suspected terrorists genuinely terrifies me.

User avatar
Dream
WINNING
Posts: 4338
Joined: Tue Dec 04, 2007 7:20 pm UTC
Location: The Hollow Scene Epic

Re: (computer) Music at its most basic.

Postby Dream » Wed Oct 13, 2010 6:11 pm UTC

Роберт wrote:Why don't you fix the Wikipedia article I linked to then, since it's apparently wrong.

Not everyone in the world cares whether Wikipedia is right or wrong. I have a special extra layer of not caring for times when right and wrong hinge on precise definitions of terms, both of which are accurate by their own standards.

In DSP terms, compressing a waveform without changing its relative shape is (very strictly speaking) distortion. The wave started out one shape, and after you did something, it was a different shape. Since you can't have an absolutely perfect interpolation even at 32 bit floating point precision, some distortion proper will exist too. The fact that that shape being the-same-but-quieter is called distortion in one discipline, and attenuation in another is not important to anyone but lexicographers. The sound (tone, timbre, whatever you like) starts and finishes same in both cases. In your system the timbre of the sound would change, as the relative levels of peak amplitudes within the waveform (which is what makes up the timbre) would change. That isn't limiting as it is understood in an acoustic sense, it's synthesis.

If you want to get your head around this, you're going to have to understand the difference between acting on individual cycles within a waveform and acting collectively on audible sounds made up of thousands of those cycles. Your per individual sample approach is a synthesis technique, and even if you can call it limiting, it's not the kind of limiting you use to attenuate a whole sound. Insisting on this is as pedantic as insisting on calling a guitar a percussion instrument because picking is technically striking the string with a pick.

Also, telling me that if I like my definition of limiting so much then I should go marry wikipedia is not going to extend my patience very far. Try to remember that you're not an expert on this, and that your tone should probably be more pleasant. I know several people who could school me on DSP and acoustics, and in general I avoid rhetorically appealing to wikipedia when dealing with them.
I knew a woman once, but she died soon after.

Роберт
Posts: 4285
Joined: Wed May 14, 2008 1:56 am UTC

Re: (computer) Music at its most basic.

Postby Роберт » Wed Oct 13, 2010 6:46 pm UTC

First of all, I'm sorry my tone came across wrong. I was merely trying to point out that the definition I had in my head of limiting was a valid one.

Dream wrote:If you want to get your head around this, you're going to have to understand the difference between acting on individual cycles within a waveform and acting collectively on audible sounds made up of thousands of those cycles. Your per individual sample approach is a synthesis technique, and even if you can call it limiting, it's not the kind of limiting you use to attenuate a whole sound. Insisting on this is as pedantic as insisting on calling a guitar a percussion instrument because picking is technically striking the string with a pick.


For the record: I have a BS in computer engineering and a minor in math; I encounter DSP stuff occasionally in my job as a software engineer. I do understand the difference between instantaneous voltage and RMS power etc.

I am not insisting that limiting can refer, in the digital realm, to per sample without out a reason. Let me quote the exact same item I did before:
Soft limiting is limiting in which the transfer function of a device is a function of its instantaneous or integrated output level.

Here's another definition:
a nonlinear electronic circuit whose output is limited in amplitude; used to limit the instantaneous amplitude of a waveform (to clip off the peaks of a waveform)) "a limiter introduces amplitude distortion"

Instantaneous. Digitally, that would be per sample. The guitar=percussion analogy is flawed, because when I ask Google what a limiter is, its first response tells me it is instantaneous, not with the attack/release paradigm of a compressor.
When I ask Google what a guitar is, the first results say nothing of percussion. Rather, they call it a stringed instrument or a musical instrument.
The Great Hippo wrote:[T]he way we treat suspected terrorists genuinely terrifies me.

User avatar
Midnight
Posts: 2170
Joined: Mon Dec 10, 2007 3:53 am UTC
Location: Twixt hither and thither. Ergo, Jupiter.

Re: (computer) Music at its most basic.

Postby Midnight » Wed Oct 13, 2010 10:23 pm UTC

Children, please.
Dream wrote:The difference may not be noticeable, but the interference patterns it generates will be. So while you can quite happily ignore the difference between 440 and 440.01, you may have to stop ignoring it when you begin to stack many waves that will give rise to unexpected frequency relationships between their overtones. Creatively you're on solid ground, just do what sounds good. Technically, it's probably worth giving yourself a mechanism to retune as necessary to eliminate unwanted beating.

...which is why I'm using just temperament instead of equal. Using math to get a theoretically perfect E = 330hz is better than any other system, because all the other systems, modern-western-equal-temperament included, are based upon compromise. In lieu of using frets and valves and such (fretless basses whoo) for some semblance at equal temperament, I can be perfect, which means no weird beat things unless I want them to. Four part chords of 165/330/412.5/495 (E/E/Ab/B) sound fine, whereas (using that Michigan University thing) 164.81/329.63/415.3/493.88 sounds all kinds of wonky. WHICH COULD BE USEFUL LATER, and I was thinking that I'd switch the chords from Just to Equal later to make the sonic chaos a little bit more chaotic. but for now, no thanks.

Tonight I'll enter the midi lab and fuck shit up like nobody's business. Any ideas for interesting things to do in reason.. 3? I think we're still using 3. I figure I'll use reason as a plugin for pro tools, where I shall apply mad soft limiting and some actually good delays and reverbs (hooray Lexicon!), and then see about sampling pieces of the songs and putting it other places, BT-style.
uhhhh fuck.

User avatar
Dream
WINNING
Posts: 4338
Joined: Tue Dec 04, 2007 7:20 pm UTC
Location: The Hollow Scene Epic

Re: (computer) Music at its most basic.

Postby Dream » Thu Oct 14, 2010 12:23 am UTC

Midnight wrote:Four part chords of 165/330/412.5/495 (E/E/Ab/B) sound fine, whereas (using that Michigan University thing) 164.81/329.63/415.3/493.88 sounds all kinds of wonky. WHICH COULD BE USEFUL LATER, and I was thinking that I'd switch the chords from Just to Equal later to make the sonic chaos a little bit more chaotic. but for now, no thanks.

As long as everything is happening because you want it to, and you're aware of how to unhappen it if it stops being wanted, rock on.

Midnight wrote:Any ideas for interesting things to do in reason.. 3? I think we're still using 3. I figure I'll use reason as a plugin for pro tools, where I shall apply mad soft limiting and some actually good delays and reverbs (hooray Lexicon!),

Is it a Lexicon hardware unit? Which one? If it's a PCM 96 or 92, check out the resonators for a good (equally tempered) way to generate chords and overtones. They sound totally lush. I don't think the software version of those include the non-verb/delay effects, though I could be wrong.

In Reason, the reverb has a built in gate, which can generate very cool effects when fed with noise in short bursts. I'd also be looking at the CV patching on the back panel, using the Spider units to drive loads of parameters on the synthesizers from various oscillators. If I recall correctly, there's a CV merger, which is cool for creating non-sinusoidal modulation waveforms, which are great.

Midnight wrote:sampling pieces of the songs and putting it other places, BT-style.

Find the transient that makes up the start of a syllable. Highlight the 30-60ms after the transient finishes, make it a region, and loop it as much as you need to. Overwrite the original region to make sure the timing stays where it's supposed to. Don't loop so far that you obscure the following syllable. This keeps the form of the word while giving that buzzing edit thing. Now I'm going to go clean the auto-tune out of my ears...

Роберт wrote:I do understand the difference between instantaneous voltage and RMS power etc.

I know you understand the difference. But you yourself are talking about managing overall output levels by limiting, and working with sub-wavelength sample values is not limiting in that sense. You're not going to effectively manage output levels with waveshaping unless you basically just lower the overall volume. If you start applying progressive response with respect to increasing intensity, you will get distortion. I know because I've used it to do just that. You can even write functions that will distort into useful harmonic series, the Chebychev functions. But if you want distortion free limiting, you simply have to work in a larger scope than the waveform of the sound, that's all there is to it.
I knew a woman once, but she died soon after.

Роберт
Posts: 4285
Joined: Wed May 14, 2008 1:56 am UTC

Re: (computer) Music at its most basic.

Postby Роберт » Thu Oct 14, 2010 2:34 pm UTC

Sounds cool, Midnight, and Dream's suggestions sound awesome. Are you going to share some of the results of your experimentation with us?
The Great Hippo wrote:[T]he way we treat suspected terrorists genuinely terrifies me.

User avatar
psykx
Posts: 408
Joined: Sat Feb 23, 2008 11:24 pm UTC
Location: England
Contact:

Re: (computer) Music at its most basic.

Postby psykx » Fri Oct 15, 2010 10:04 am UTC

I'm pretty new to music theory and I'm mostly self taught, so apologies if I've missed something basic. But is there a universal rule which means notes will sound good together? Obviously note frequency differ on each instrument. Are cords different of each instrument? or do the frequencies not differ enough? Is there some kind of mathematical relationship which would predict (particularly amazing) harmonies between particular instruments?

edit: @Dream, I am actually planning to respond to your PM but I want to do so properly and I keep remembering when I'm at work and can't listen to music from the computer.
Berengal wrote:Only if they're killer robots. Legos are happy robots. Besides, even if they were killer robots it wouldn't stop me. You can't stop science and all that.

Роберт
Posts: 4285
Joined: Wed May 14, 2008 1:56 am UTC

Re: (computer) Music at its most basic.

Postby Роберт » Fri Oct 15, 2010 2:33 pm UTC

psykx wrote:I'm pretty new to music theory and I'm mostly self taught, so apologies if I've missed something basic. But is there a universal rule which means notes will sound good together?

Sounding "good" is subjective, but generally notes whose frequencies have a low integer ratio to each other work really well together. A "perfect fifth" is a 2:3 ratio, so the a note at 440Hz would have a perfect fifth at 660Hz. If it is close to that, it will still sound good. Intervals that don't have low integer ratios tend to sound dissonant, and the amount of acceptable dissonance is a matter of taste, but generally you have to put dissonance in a context that makes the listener ready for it and resolves it for it to sound good. A "dominant seventh" (If A is the root note, the dominant seventh would be G) is an example of a dissonant sounding interval that is used frequently in music.


psykx wrote:Obviously note frequency differ on each instrument. Are cords different of each instrument? or do the frequencies not differ enough? Is there some kind of mathematical relationship which would predict (particularly amazing) harmonies between particular instruments?

...?
Generally, when playing instruments together, you want them all tuned the same. Most modern western instruments are tuned to "concert pitch" which has the A tuned to 440Hz, and uses equal temperament tuning system, which means (simplified) your ratios for fifth and fourths and other intervals won't be perfect, but you can play in any key sounding equally good. So all the notes will be the same frequency from instrument to instrument. Does that answer your question?

Really, I think you should continue your study of music theory, because there's a lot there, and it's very interesting stuff. Good luck!
The Great Hippo wrote:[T]he way we treat suspected terrorists genuinely terrifies me.

User avatar
Midnight
Posts: 2170
Joined: Mon Dec 10, 2007 3:53 am UTC
Location: Twixt hither and thither. Ergo, Jupiter.

Re: (computer) Music at its most basic.

Postby Midnight » Sat Oct 16, 2010 2:03 am UTC

psykx wrote:I'm pretty new to music theory and I'm mostly self taught, so apologies if I've missed something basic. But is there a universal rule which means notes will sound good together?

Well, in Western music, there are scales for that sort of thing, and the two most important ones are the major and the minor. Major/minor is differed by the 3rd note in the scale (so, if your root is C, the 3rd is E); a major third sounds nice and happy, a minor third sounds ominous (minor 3rd = flat third. Thus, C and Eb).
"Perfect" intervals (the 4th and the 5th--F and G, if our root is C) also sound good, unless you mess with them by augmenting/diminishing them (sharping/flatting). Octaves always sound good together cause they're the same note, just a different pitch.
psykx wrote:Obviously note frequency differ on each instrument.

Not really. You can play a C on any western-tuned instrument (by western, I mean All The Popular Music You Hear Today). Some instruments have different ranges--a bass can hit lower notes than a flute. A piano can hit all the notes. There's a solid crossover between bass and guitar, as well. The main thing that differs one instrument from another is timbre. I'm sure someone else can give a better introduction on how to generate a specific timbre using just a computer, but it boils down to this: instruments are really complicated and there's a helluva lot going on, acoustically. There's overtones, there's buzzes, distortions, chorus effects, partials, and so forth. That's why it takes quite a bit of time to get a usable trumpet sound just through FM synthesis or additive/subtractive synthesis, and there are people paid to do that sort of thing, so you're better off buying a synth instead of spending sixth months to get your synth to sound like a horn.
psykx wrote:Are cords different of each instrument?

Depends. You can't play chords on a wind instrument, and a bass has less strings than a guitar, so it requires some finagling to play guitar chords. But you can play an E major chord on a bass, or on a guitar, or on a piano. They all use the same tuning system--equal temperament--so they all hit the same notes, so they can all hit the same chords.
psykx wrote:s there some kind of mathematical relationship which would predict (particularly amazing) harmonies between particular instruments?

I dunno what you mean by particular instruments, but the frequency of a major third is 1.25x the frequency of its root note, a perfect fourth is 1.333x its root, and a perfect fifth is 1.5x the root. So, an A is a sine wave beating at 440 times per second, and an E above A (the perfect fifth) has a period of 660hz.
Even though accurately reproducing a particular instrument's timbre is difficult, it does boil down to sine waves. The period (width of the wave, one cycle of up and down--a period of 440 means the wave goes up and down 440 times per second) determines the pitch of the tone and the amplitude (how high the wave goes) determines the volume.
uhhhh fuck.

User avatar
icelizarrd
Posts: 53
Joined: Sat Nov 14, 2009 10:15 am UTC
Contact:

Re: (computer) Music at its most basic.

Postby icelizarrd » Sun Oct 17, 2010 5:08 am UTC

This is a fascinating topic to me: I don't think I've spent enough time delving into low-level synthesis myself.

Dream, do you think Supercollider is easier to pick up than CSound for this sort of project? I was just wondering why you mentioned one and not the other. (There are others, of course, but CSound's such a widely-used option.)

psykx, do you mean that the overtones for each instrument differ? Because if you're just talking about the fundamental pitch, then yeah, there's no reason you can't tune instruments to (or design them to be) the same frequency. (On the other hand, it's true that some instruments' tuning can't feasibly be changed "on the fly"; so in that sense, a violin can play a chord that a piano cannot precisely replicate, if the piano has not been tuned properly. Additionally, certain wind instruments produce slightly flattened and sharpened notes depending on the valve fingering; but this can be compensated for.)

I don't know that anyone has come up with any theories about the interaction of specific instruments' timbres; but people have certainly tried to categorize "good sounding" harmonies according to their ratios. Historically, it has often boiled down to "the simpler the better" vis-a-vis ratios. This link details some theories about how this works.

User avatar
Dream
WINNING
Posts: 4338
Joined: Tue Dec 04, 2007 7:20 pm UTC
Location: The Hollow Scene Epic

Re: (computer) Music at its most basic.

Postby Dream » Sun Oct 17, 2010 11:25 am UTC

icelizarrd wrote:Dream, do you think Supercollider is easier to pick up than CSound for this sort of project? I was just wondering why you mentioned one and not the other. (There are others, of course, but CSound's such a widely-used option.)

CSound is a non-real time environment. I haven't used it, but I have used other "offline" systems, and they are great for either batch processing or entire compositions, but it is very useful to have a very solid idea of what you are trying to achieve, and how you are going to go about it before you start. You have to write the entire program from start to finish, compile, bug-fix, compile again, render, usually wait a while, then listen, make changes and repeat. Supercollider on the other hand is a real time system in which the programming language runs separately to the real time audio engine, sending it commands rather than rendering the audio itself. So you can have the sound playing as you compile new chunks of code individually, send them to the engine and hear the results immediately.

The immediacy and responsiveness is what I'd say SC has in its favour over CSound. Far better for learning, because you can learn and do a single thing at a time rather than having to get it all right, right from the start. I can recommend Common Lisp Music as an offline environment, because the whole list based concept lends itself to music very well. The elegance of the (programming logic) loops within loops comprising the entire program, which comprises the entire piece of music, is a very satisfying thing.
I knew a woman once, but she died soon after.

User avatar
Ivor Zozz
Posts: 169
Joined: Sun May 09, 2010 7:35 pm UTC

Re: (computer) Music at its most basic.

Postby Ivor Zozz » Mon Oct 18, 2010 10:37 pm UTC

I've tried both CSound and Supercollider and prefer SC for its immediacy and compact expressions (CSound is quite verbose). I'm not extensively experienced with either, though.

You could also try out ChucK, which is pretty easy to learn, but seems (in my limited experience with this stuff) a bit less powerful than SC.

Excellent thread!
The Classical I call the healthy, the Romantic the sick. - Goethe

User avatar
icelizarrd
Posts: 53
Joined: Sat Nov 14, 2009 10:15 am UTC
Contact:

Re: (computer) Music at its most basic.

Postby icelizarrd » Tue Oct 19, 2010 11:33 am UTC

Ah, I see, thanks for explaining, Dream.

I'm giving SC a shot now, and I do find it more rewarding to work with than CSound*, even though I think I like some of CSound's conventions a little more. SC's OOP approach is pretty nice.



* Not, by the way, that I ever got very far in my CSound excursions either.


Return to “Music”

Who is online

Users browsing this forum: No registered users and 9 guests