h2g2 The Hitchhiker's Guide to the Galaxy: Earth Edition

Hi Jerms, it's been a while but I remember talking to you on a few threads when you were around before.... a couple of years ago.

I was glad to find you've returned, had a scan of your journal and I'm very interested in your music generation research. (I make electronic music but am definitely NOT a programmer!)

Anyway, if you fancy a chat, I'm here

Yay! Great to hear from you!

The music generation is an idea I've had floating around for a while... I tend to attempt a fruitful idea until I hit a major obstacle, and then give up for a month or so... From my research the other day I got a couple of ideas for some algorithms I'd like to try... would you be able to help me with the theory perhaps? I'd certainly appreciate any ideas you have. I'm a programmer, but not much of a musician!

What types of music are you most into? What's your forte when you perform?

Catch you 'round!

Hi

I hope you don't mind, but your question has provoked a lot of thought and I've put it all here... I hope it doesn't seem overwhelming.



I like a wide range of music to listen to...the music I'm into making at the moment is experimental electronica. I like to use software in different ways - to make sounds that are not necessarily what the programmer originally intended.



For me, music could be defined as "sounds that are stimulating" - I think music can be anything that creates an emotional, physical or mental response in an individual (or group) that hears it...

So, it follows that I class most sound as music, but if I had to narrow it down I would add that it would have to be sound made intentionally by some living being, usually human (so human speech would be included), but I might also include bird song, dog barks, crickets chirping...

I create music with this definition in mind, but that doesn't stop me from using melody and rhythm...it just means that I use lots of other sounds and see them as equally important. To me, the proof of the pudding is in the eating, ie, if the resulting sounds provoke the aforementined types of response, then it must be music!



If you have the time and the inclination, you can read my musical biog here:

http://www.candinook.cjb.net

unfortunately, the mp3s don't work at that site, but if you want to get an idea of what some of my music sounds like, there are a couple of snippets from my latest release here:

http://www.fencingflatworm.connectfree.co.uk/ff023.htm

....but I don't want you to feel pressured!

I really am interested in your project and would like to help if I can. What aspect of the theory do you need help with? My maths is a little rusty but I do know a fair amount of musical/sound theory (or at least know how/where to look it up!)



Oh, and I would be happy to talk about any other subject too - I am not only interested in music...

Fantastic!
Unfortunately the premises I want to ask your opinion on, are running through my head too quickly for me to type them out as they happen.... let's see how many I can catch on the way past, in no particular order.

One of the purposes of the sofware I would write, would be that it could be used in ways it wasn't intended - more on that in a minute.

"Sounds that are stimulating" - I'll agree with that. But hang on... physical or mental response? You've also left out the possibility of negative responses to the music. Obviously one doesn't have to have a positive response in order to class its cause as music; plenty of people dislike particular songs. And what about the physical response of an eight hertz tone; just because it has a laxitive effect doesn't mean it's music, right?

and "sound made intentionally by some living being..." What about the soothing sound of a train? Most would agree it's quite musical. Come to think of it, I'm not sure about intentionality either... what about Marvin and the amazing muppophone? (or is it a muppet-phone? I can't remember.) I suppose the intentionality then is in Marvin, holding the rubber hammer, and not in the creatures making the sound.

....

On reflection, I think the definition of music, as it differs from sound, would have to be closer to an expression in terms of a pattern - humans are, by default, creatures of pattern. This is why we learn by repetition, this is why we seek pattern in our lives, our music, in the labelling of the people we meet for the first time. Everything.
Music always repeats; there has to be some kind of repetition for it to transcend mere sound. Do you agree?
Factors such as the speed of its repetition play a large factor in the responses they produce - among a /huge/ number of other things! If the sound repeats slowly it becomes relaxing, I'm guessing our brain slows down because it takes less brainpower to 'figure out' the pattern when it has more time. Conversely, if the tempo is too quick, then we become more active. But there are limits each way; If a tap drips we can find it musical, but if it drips too slowly it's just annoying; perhaps we lose the beat. Also I've heard some pretty extreme Acid White-noise Punk tracks which are on my limit of the difference between music and static.

And the proof of the pudding is definitely in the eating!

So lets say I write a program which attempts to create music in some way; how do I do it? (this is me thinking aloud here, of course! ) Well, the useful algorithms I've seen recently include the layered natural wave - probably the best option at the moment; it ends up looking like a biorythm graph, but in several dimensions (time, pitch, volume, opacity, etc.), timed so that the beat points for all the music objects (think of instruments or samples if it's easier) are all at the same time (although they don't /have/ to be at the same time though. Hmm.)
I've also heard of an interesting idea where three seperate modules of a program: create some random noise; 'listen' to it, using a genetic 'ear' algorithm; and reject the non-melodic parts of the creation, restructuring it into what is aparrantly quite good music. The downfalls of this one are that it would take quite some time to code the relevant parts; it would take about as long again to train the 'ear'; and the music can't be played in real-time, so would therefore be pointless in a responsive system. I like the idea of the 'ear', though.
Yet another idea with merit is for the learning/emulation system which listens to different pieces in a particular genre or by a particular artist, and tries to copy it in a general way. Aparrantly some dude has written Mozart's 42nd symphony that way.
Again, that has some appeal, but it's not quick enough to use in a responsive generation system.
And without trying them myself there's no guarantee that any of these ideas would come up with music of a reasonable quality. It's hard enough for musicians to say what it is, exactly, which makes something tuneful - let alone the poor guy trying to define that in code!

I guess I'll just have to try it; I think the main problem would be in the details - my first success would be if an algorithm can come up with a simple tone which changes according to a chaotic pattern - like 4/4 time, four notes to a bar, when the first three notes are the same (or at least linear) and the fourth breaks the pattern - becomes louder, more complicated, rises in pitch... something. Maybe even all of those. And then the same notes happen for the next two bars too, but the fourth bar changes in the same way the fourth note did, just on a larger scale. And then the same thing happens with the sixteenth bar, again.
You get the idea; and that would be a good start.

Um. Any thoughts?

Please forgive me. I just wrote a detailed reply to your last post and it was lost by some ing error when I tried to post it.

Just thought I'd post this while I compose another detailed reply and try to remember everything I said...

Oh dear god I hate it when that happens.



I know your pain: for a while there it seemed that every e-mail I tried to write would die in transit... I'd attempt it four or five times before storming off in a huff - it was either that or throw the computer out the nearest window.
And then people started complaining that I never kept in touch. Harumph!

my interputation is "Music" Will have Rythm Between the beats.
melodys are a interaction of the rythm of different beats.
as all beats are only vibrations,its not supriseing to hear
the harmonies of hot trainwheels on cold tracks,with the underside of the train acting as a big sound box.

Jerms, thanks for your patience and understanding

Here’s the re-thought out, re-written reply (I typed it out in Word first and saved it just in case, this time!)– sorry it’s taken me so long….

What I should have said in my definition was "emotionally or mentally stimulating" – I didn’t mean sounds that have a purely physical effect….

But as for leaving out negative responses – I didn’t make a point of mentioning them, but believe me, I think they’re just as important as positive ones. Of course it’s all a matter of opinion – what some find negative, others find positive; also, one can respond negatively to something, then later realise that on balance, the experience was a positive one. (I used to be a noise musician, see! Check out the Susan Lawly release “Extreme Music From Women” http://www.emfw.freeuk.com/ )

Your point that “intention by a living being” is not necessary is a good one – the train noise is a good example.

But I disagree with the notion that music has to include pattern – there are some soundscapes, ambient music and noise music that have no discernable pattern. I am tempted to make a comparison here with abstract visual art - maybe that would be a mistake, but I hope you see my point.

However, for the purposes of your program, I can see that a pattern is necessary. The idea you discussed of a simple tone changing on the fourth beat or bar sounds interesting to me.

One thing I would like to know is how much interactivity would be possible? Would the program just work on its own or would there be room for human input too? Obviously as a musician I’m biased towards there being a little interactivity – then the resulting music would be a collaborative composition between the program, the programmer and the musician using the program – I like that idea, but it would be interesting either way…..
Please correct me if I’ve got the wrong idea about what you’re trying to do.

"...it’s all a matter of opinion – what some find negative, others find positive; also, one can respond negatively to something, then later realise that on balance, the experience was a positive one." Yes! I agree completely with all of that!

Haven't yet had the chance to have more than a skim of that last link you gave me; I hope to read it more thouroghly tomorrow. (I would say tonight, but I finish work at midnight so I'm not sure if I would be able to.)

Getting deeper into it now, and I find that the definition of a pattern is a tricky one too...
I'm not adament on this, but I personally believe that even soundscapes, ambient music, and noise music have a pattern to them. Note: I didn't say it would be a /discernable/ pattern (neither did you, I notice! ), just a pattern. Maybe we subconsiously recognise it as a pattern, but consiously are unable to follow it. Or something.
An easy example I can think of would be of layered harmonics; the image I have is of those creatures in the movie "the Dark Crystal" - I can't remember what they're called but one of them is the "Master" of the main character, and they have an interesting 'hum' thing happening. One starts humming a constant tone, and another joins in with a different tone, and another, and another. Sort of like a barbershop quartet, in a way, but the interesting thing is that the different tones are all in harmony. The question in terms of coming up with a program to emulate that, is of /which tones to use/? obviously one can't just add several random tones together and expect them to be harmonic, so is there some rule for which tones to use? I've heard that there's a mathematical formula for working out at which frequencies natural tones occur (like those produced on a well-tuned piano), but I don't know what that formula would be.
Come to think of it, even one constant tone still has a pattern to it - the repeating vibration of the soundwave itself.

And as for the program's interactivity... Well I guess now is a good time to express some of the ideas I've had for the use of it; I've been putting this off for an appropriate point in the conversation. Do you remember I mentioned that I was hoping for the program to be used in ways I hadn't imagined?

I had the original idea for the program while I was playing a computer strategy game. I can't remember which one, but let's say it was Diablo 2.
Basically, it was one of those games where you (the main character) run around various levels; exploring dungeons and ruins and villages and so on; killing monsters and collecting the money and objects they drop, and using them to make your character stronger and more powerful and better able to kill stronger monsters.
One issue I'd always had with these games is that the backgound music gets very repetitive very quickly. Obviously it would be a much more playable game if the music responded to the action onscreen, in the same way that an orchestral score responds to the action on a movie screen (think of the Lord of the Rings trilogy, here).
Thinking about it more, I realised that there could be many factors influencing the outcome of the music played, even in realtime (or just-in-time play, which is almost the same as realtime, but slightly more versatile.) Some of these factors could be: the amount of enemies onscreen, the type of weapon or spell or whatever the main character is using at the time, the terrain in the area, even the amount of fog could 'dull' or 'fade' the music... the sound could even reflect off buildings using some kind off echo, based on the distance of the main character from those buildings, perhaps.
Also, sometimes in the game there are 'secondary characters' - in the case of Diablo 2, these are rangers you can hire to fight with you, which the computer controls and who walk around the levels with you until they die. I thought perhaps there could be a main riff, or theme tune or something, for the main character, and secondary tunes for secondary characters, etc.
Extending that idea, there could be tunes for different types of monsters or whatever as well, and maybe have the tune get quieter or decrease its timbre or something as you lose your health.
Also it would be kinda cool if the sound effects could sync in with the music, but wether that would make the game more or less playable would remain to be seen.

Thinking about the concept more, I quickly realised that many types of game would benefit from this kind of responsive music-generation system - the next example I thought of was for first-person shooters (think of Doom, Quake, Unreal tournament, Wolfenstein ET, etc.)
It wouldn't take much adjustment from that first type of game to the second, as many of the concepts are still in place; health, monsters, terrain types, buildings, type of weapon, etc. The only major difference is that the first type of game has more detail in it, and the second is more active.
Thinking about it more, again, I realised the best way to design the program would be as a third-party add-on to the game; I could sell the game developers a complete application for the development of the music, and they could basically 'plug it in' by telling the application information about each object (the player, the monsters, the buildings, whatever they wanted) and letting the generation application come up with the music using that information. As long as the information was continuously up-to-date, it should work quite well.
Also, since its the developers themselves who would decide which information would be passed to the app, by necessity they wouldn't be restrained at all in what they passed. So they could use it for /anything/ and the app wouldn't care; the only issue would be how to quantify the information the app is passed.

So, by definition, such a program could be used in almost any way; I had visions of dancers on a stage, with IR detector dots on their bodies, dancing along to the music which was being created based on the movements of their dance, as it unfolded. Or someone with a Virtual Reality headset and gloves, playing music on a set of virtual instruments by adding layer upon layer of sound from each instrument.

Anything is possible.

So in response to your question on the amount of human input... the answer would be "as much input as you want". I hope.

Phew.
What do you think?

That sounds brilliant - I don't play computer games but at times when I've had a go the thought had occurred to me "wouldn't it be good if the music/sound could interact with the player as well as the visuals"
It would enhance the experience so much.

The dancers idea is weird but fascinating - very appealing - and it wouldn't have to be just dancers - it could be workers on a building site or anything!
Not sure what you mean by the "someone with a Virtual Reality headset and gloves, playing music on a set of virtual instruments by adding layer upon layer of sound from each instrument" thing though - why the virtual reality headset? What would that acheive that just real keyboards can already do? After all, we still only have two hands, even in virtual reality we couldn't get round that...or could we? Hmmm I'll have to think more about that one!


(By the way, you're right about pattern - of course the very fact that sound is a wave - this is reinforced when we feel the vibration of very deep bass sounds that are just below our range of hearing...)

Workers on a building site? Interesting... I suppose anything really does mean 'anything'! Schools of fish, or temperature fluctuations in a fire or anything...

The VR idea is a bit hard for me to explain without being able to flap my hands around... basically, if you imagine a virtual bank of different instruments in front of you, sort of holographic-looking... and pointing at an instument makes it fly into your hands, or line up in front of you or whatever... you can then 'play' that virtual instrument... the program would be able to work out the tune after a couple of repetitions... at which point some visual object, maybe a small geometric shape like a cube, appears hovering in midair just at the top of your field of view, slightly in front of you, glowing the colours of an E.Q. in time to the music.
The cool bit now is that even if you stop playing the instrument, the tune will continue, allowing you to do the same with other instruments - build up several layers of sound, and a collection of those visual objects to go with them.
Because those objects are representations of the various tunes, of course it would be possible to pick them up, prod them, move them around, throw them away, whatever... and the music would respond accordingly. Move it to the left, its tune would drop in pitch. Move to the right, the pitch would increase. Move it down and the volume would drop, up and the volume would rise. Forward and back would be for... um. Tempo perhaps? Fade in and out? I'm not sure.
Tap it to make it change the tune slightly, double-tap to pause or mute it, and crush it in your fist to make it stop completely.
I hope you get the idea. And that's just the possibilities using VR gloves and headset; an entire VR bodysuit would open up a world of rhythm possibilities! Imagine the dudes from Stomp, or the Blue Man Group, all decked out this way!

theres already a program like that without the graphics
you've described,or what i think youve described.
()Onion machine()or some thing like that ,US school site something luke that searched for it cant find it.again, java would be usefull for hologr angle.also windows touch screen ,soon to be announced,
be handy

Actually thinking about it more, I've changed my mind on what would be my first success in writing the program; it would be something even simpler than I'd thought - either layered harmonics, like I mentioned before; or something where the tones are preset but the timing is important, like playing African drums.

try http://spanky.triumf.ca/www/fractint/whats_new.html
http://spanky.triumf.ca/www/fractint/fractint.html
came from 2003 thread hear

Fantastic! Thanks, dude!

Any idea why VR died? I heard it 'got too far advanced for its own survival' and it 'hit an evolutionary dead-end', but I have no idea why.
Suddenly nobody seems to care, or to think it's feasable...?

doubt if its dead, bet you millitary applications
or leisure,by that i mean "flight simulaters and like of" holding up
commercialistion for joe public.along with windows ect

ah. Makes sense.
Last I heard the only major troubles they had was that 1) the computers they were using were too slow, which shouldn't be a problem any more; and 2) when anybody tried to walk anywhere they would walk into walls they couldn't see, and any way of restraining them would spoil the illusion - one idea was a small round platform, ball bearings for the floor, padded rails all around, and harness to keep you in the middle. It almost worked, but aparrantly you could feel the harness tugging on you.

I was talking about this to my flatmate, and he thought a good way around this would be to have the system exclosed in a sensory-deprivation tank. I'm not sure if that would work; again the physical pressure in various parts of the body wouldn't match what the people think they should be feeling; and there's also the issue of safety - if someone tries to turn around and discovers they're face down in a pool of saline, they might panic and drown themselves.
Then again, a responsive bodysuit with built in ventilation and so on would combat these problems.
What do you think?

lateral thinking best "harness" wind tunnel
depending on "game" vertical horizontal tube,platform.
computer controlled up down on board (builtin scales)
or remember wheel of death,centrufugal force techno
could contain answers.

hmmm... good ideas.

I guess the only 'perfect' solution would be that of the matrix - plug in something to short-circuit all external sensory input, and replace it with virtual input. Anything else and the best you can do is to disguise it instead.

...still thinking about your ideas... I'll get back to you.

plug in something to short-circuit all external sensory input.
we have that ¬party political broadcast¬
Forget tunnel .Large Ball held like hamster ball.
Sensors could be placed on outside of ball,to take positions of
"users" screen technology could enable project images straight on to retina via "laser" enabling dispensation of dreadfull headsets.