PDA

View Full Version : Does the future of Music lie with the Midi Interface?



Woebot
15-10-2004, 01:39 PM
Quite often these days one comes across websites which produce, via the the programming in one's one computer's sound card and BIOS funny tootling music. What's happening, and correct me if I'm wrong, is that in the same way HTML functions and vector graphics documents work, a few integers communicate to one's computer what noises it should make.

Presuming computers and other such devices do continue to have enlarged resources at their own disposal isn't it fair to assume that music will be transported in the future in much smaller digital packets. I mean for instance the mp3 is essentially a bitmap, snapshot of a recording. It's funny to think that this is closer (if not identical) to the historic principle of sheet music, itself a "programme" rather than an encapsulation.

On the other hand I guess that the increasing speed of mainframes and the enlargement of the trunk roads of data communication (from 3G mobile phones to broadband) may create a situation counter to this whereby in fact we all end up downloading AIFFs.

I'm sure there must be some top theorising along these lines.

hint
15-10-2004, 02:14 PM
What you're referring to is General MIDI.

This is an ageing standard soundset that was established to enable the kind of process that you describe as well as music in early videogames machines etc. MIDI information is extremely simple and therefore can be transmitted over the net extrememly quickly. Likewise, it would have taken up very little precious space on games cartridges or cassettes for the early consoles and games computers.

The limitations of this practise come into play when the MIDI information is received and processed.

On my Mac, I have Quicktime player. This incorporates Quicktime Synth - a very simple synth that contains the General MIDI soundset. So when my I go to a page that transmits a MIDI file to my Mac, the Quicktime Synth kicks in and plays the tune.

However, this relies on an established standard. The composer of the piece of music / MIDI file knows that if he programs a sequence on MIDI channel XX it will be received and understood by a General MIDI device and the resulting noise will be whatever sound is assigned to MIDI channel XX.

So what it boils down to is that the sonic pallette will always be limited to whichever sounds are included in your GM MIDI playback device. And these sounds have to conform to a certain standard, because otherwise the composer might write a nice piano line on MIDI channel YY and find that when it plays on your computer is comes out as a snare drum.

From a compositional point of view it's a limitless and versatile way to write music and transmit it in seconds to another computer - you decide which notes are played, how loud they are and which sound to use, just like writing sheet music for an orchestra (as you point out).

From a sonic / sound design point of view, it's naff - the sounds at your disposal are, on the whole, dire. So it's more like writing for an orchestra of Casio keyboards I suppose. Whilst these sounds individually can be interesting and useful (the Casio VLTone line in Outkast's Hey Ya proves this), it's been a while since I heard a piece of music performed entirely via GM MIDI on a commercial release!

Hope I managed to explain it correctly.

Woebot
15-10-2004, 02:59 PM
thanks :-)

i guess what im positing is the idea that maybe in the future (when we're all driving hovercrafts :p) a greater barrage of sounds ends up being stored at the recieving end, thus opening up the music beyond the (as you rightly point out) naff palatte already available. i suppose this is the principle at the heart of the dev of vectors, that the points of reception have grown more sophisticated, enabling whats being transmitted to become more developed)

be.jazz
15-10-2004, 03:50 PM
What's the point of it, though, if one can download the "bitmap" in seconds?

Woebot
15-10-2004, 03:58 PM
ay you got me there mwanji! just kicking the idea around.

mainly i suppose cos theres no reason it wouldnt be possible.

be.jazz
15-10-2004, 04:03 PM
Maybe if someone wanted to write "sheet music" for computers that individual users (performers) would interpret using their own, home-made sound palette? I'm kind of scraping the bottom of my shallow barrel of imagination here... ;)

DigitalDjigit
15-10-2004, 04:05 PM
My first response was "no, this will never happen". There are so many sounds out there that there is no hope to store them all. However!

Imagine a program like Reactor or or MAX/DSP or something where you have basic building blocks like oscilators, filters etc. at your disposal. You can store all the settings and programming the composer makes on his end and send them over thus recreating the recording more or less precisely on the other end.

The question is then: can a finite set of oscillators/filters/envelopes/what-have-yous produce any sound? I think the answer is no. There will always be a sound that a given set-up cannot produce. But since most producers use a fairly limited sound palette (when compared to infinity :) ) this could be sufficient.

Any sound in the universe can be deconstructed into sine waves (oscillators) however the number of sine waves required may be infinite and you can only get succesively better approximations with a finite number of sine waves. This is what mp3's do if I am not mistaken. So maybe it is possible to create a sound so complex that a file containing all the programming necessary will be as large as a recording of it.

Also, is the above true for sounds with a maximum frequency contained within of 22Khz? Can a finite number of sinewaves represent any such sound? Doesn't this have something to do with the Nyquist limit? In that case will a lossless mp3-type format be possible that still compresses somewhat?

Rambler
15-10-2004, 04:52 PM
If I remember rightly from my undergrad days playing around in the university electronic studio, MIDI is really useful compositionally, and the sort of information you can transmit can be really complex (pitches, rhythms, dynamics, channels, etc etc). However the whole set up is still built around a fairly traditional model of the composer-performer relationship. With a pencil and a sheet of score paper, any composer can write musical instructions of the same sophistication as someone using MIDI can (and the advantage of analogue paper is that there are fewer restrictions on your imagination) - but as an actual musical experience, the whole thing is contingent upon getting someone to play it, and play it well. The current standard of home computer soundcards etc is roughly equivalent to having all these compositions played by a very average school orchestra. With investment you can attach all sorts of sound modules to improve the quality of you playback, and with investment and the right knowledge, you can change all the sounds so that the MIDI 'score' is played by a synthesised bongo drum orchestra, but I'm not sure how far that takes us. MIDI's great for a lot of things, but its fundamental limitation is that it still relies on the centuries-old composer-score-performer triad, and doesn't naturally open up new avenues for musical communication and experiment.

I don't know where I'm going with all this ...

MiltonParker
15-10-2004, 07:43 PM
Just a reference: Thomas Dolby spent most the late 90's pouring money and energy into a company called Beatnik, which was an attempt at a MIDI playback engine that would provide programmers with some actively interesting sounds. Unfortunately it never quite took off as a standard, and the company's retrenched it's technology largely towards the service of... ringtones

http://www.beatnik.com/products/index.html

several friends of mine worked for the company, said it was essentially fun but the stakes were too high...

Grievous Angel
15-10-2004, 08:34 PM
I guess what im positing is the idea that maybe in the future (when we're all driving hovercrafts ) a greater barrage of sounds ends up being stored at the recieving end, thus opening up the music beyond the (as you rightly point out) naff palatte already available. i suppose
Ringtons are a good jumping off point for this idea. The tech is simple: phones with polyphonic sound chips, ringtones as midi files + some DSP information for the sounds = simulations of the original tracks. There's some evidence of ringtones being used as an alternative to CDs and even as a replacement for studios. People have seen kids doing new raps over for example Usher's yeah. To refer back to Digital Digit's position that no single DSP can model any sound and therefore replicate any recording, he's right -- but ringtones don't try to do it perfectly. They can reproduce anything -- badly.

On another angle, there's a long history of music technologists trying to create formats that allow music to be released as an interactive medium. When quadrophonic and then multichannel audio formats were developed there was talk of releasing albums in multitrack format to allow users to do their own mix. Same with multimedia -- one of my CEOs used to go on about multimedia meaning listeners could "conduct" classical music.

This failed for two reasons. One was copyright problems. The other was that a lot of the value of music is in the mix and the mastering -- basically, most people can't be bothered to do it, they like having the finished article. The AIFF-quality files required would probably be a bit big for 512K broadband.

The downsides of General MIDI -- whether delivered through a browser or as a download -- have been covered thoroughly here. Commercially, it's minimal. The music publishers sell thousands of pieces of music as GM files -- mainly to MOR lounge musicians singing along to a synth backing. You can see ads for this in the back of Sound on Sound. They sell for pennies and in fairly small quantities; there's no real money in it. Even so it's likely that Microsoft and Apple will develop their GM synths so that you can do the computer equivalent of playing ringtones -- i.e. they'll stick some DSP software into their GM synths that works better and is less "toot-toot, dink-dink". This wouldn't be hard. It also wouldn't be that interesting.

However, there is a kind of equivalent of General MIDI and ringtones that works really well and is used a bit by some amateur musos. Owners of Rebirth software can download other users' songs that reply perfectly (and can be edited) on the software's 2 303s, 1 808 and 1 909. Users can also create new sound sets for Rebirth and release songs to go with them. The same is true of FruityLoops and some other soft-studios. But the results aren't that interesting.

Above all none of these interactive music formats handle vocals well if at all, unless you do your own, as I alluded to above.

be.jazz
15-10-2004, 10:23 PM
the centuries-old composer-score-performer triad
What are the new configurations?

Woebot
16-10-2004, 04:23 AM
cor. loads of interesting info here. i guess its, as well as being like a musical score (without the importance of the players interpretation, the fluidity to bring it alive) like pianola music.

obviously this might be an extremely limited music. ramblers point about the inflexibilty of the composer/score/performer triad is right, but thats the same structure at the heart of (embodied within) all electronic music isnt it. even if you hate electronic music its not exactly been a sonic apocalypse has it? also rambler its funny to reflect that in classical music the timbral varieties are in some way extremely limited, a piano has only so many keys, and yet we dont usually complain about the limitations. i know general Midi often sounds like tootling drivel, but isnt that because its being played on the equivalent of a toy piano!

fascinated to hear 2stepfan's story of people swapping rebirth and fruity loops files (same as swapping pianola reels!), and yeah i bet they sound rubbish lol. i suppose thats what you mean DigitalDigit.

really i cant think of many reasons it'd be a good idea, but (running in the face of meme's suggestion of people's lack of interest in interactivity) I guess there could be the capacity to edit and reorganise music. thinking that, much in the same way that bitmaps can be encapsulated within illustrator documents (one image can be chopped up and reused in a hundred different ways) you could attach a few "real" sounds to your "midi" file like some real recorded drum beats and have them tiled up by the files programming.

i'm sure this is going absolutely nowhere but hey!

:D

DigitalDjigit
16-10-2004, 01:22 PM
Welcome to the wonderful world of trackers! What you are describing, Woebot, is basically how tracker software works. You load samples. Then you have a pattern editor where you can write a score, so you have notes being played by the instruments (samples). The change of pitch is done simply by changing the sample rate of the sample. There are some effects you can add like vibrato but nothing like echo/reverb (at least not in the original trackers). Then you arrange these patterns (which are usually 4 bars long) and create a composition.

Since the samples are stored together with the programming you get a perfect recreation on the other end. The problem is that since the samples are uncompressed the file can take more room than the mp3 produced by it. The fact that the samples are layered on top of each other in the actual composition but have to be stored separately also saves a little space in the finished product.

There is a new crop of trackers (most notably the free Buzz, see: http://www.buzzmachines.com/ ) which allow you to add some effects to the process such as echo, distortion, filters etc.

These have been around since the late 80's, in fact some 'ardkore (Red Alert & Mike Slammer) was created with such software on the Amiga.

The problem with these is that the sound palette is very limited but wouldn't be that way if you could store the information on the building blocks of sound. So if the mp3 (pure sound) is one level and trackers are a level above (meta-sound, information on how to create sound out of samples (sound)) then my idea, which isn't all that new and revolutionary, is a meta-meta-sound (information on how to create sound that will create sound). The additional level of abstraction gives more flexibility and saves storage space.

Woebot
16-10-2004, 05:15 PM
Welcome to the wonderful world of trackers! What you are describing, Woebot, is basically how tracker software works. You load samples. Then you have a pattern editor where you can write a score, so you have notes being played by the instruments (samples). The change of pitch is done simply by changing the sample rate of the sample. There are some effects you can add like vibrato but nothing like echo/reverb (at least not in the original trackers). Then you arrange these patterns (which are usually 4 bars long) and create a composition.

Since the samples are stored together with the programming you get a perfect recreation on the other end. The problem is that since the samples are uncompressed the file can take more room than the mp3 produced by it. The fact that the samples are layered on top of each other in the actual composition but have to be stored separately also saves a little space in the finished product.

There is a new crop of trackers (most notably the free Buzz, see: http://www.buzzmachines.com/ ) which allow you to add some effects to the process such as echo, distortion, filters etc.

These have been around since the late 80's, in fact some 'ardkore (Red Alert & Mike Slammer) was created with such software on the Amiga.

The problem with these is that the sound palette is very limited but wouldn't be that way if you could store the information on the building blocks of sound. So if the mp3 (pure sound) is one level and trackers are a level above (meta-sound, information on how to create sound out of samples (sound)) then my idea, which isn't all that new and revolutionary, is a meta-meta-sound (information on how to create sound that will create sound). The additional level of abstraction gives more flexibility and saves storage space.
cool! :)

hint
17-10-2004, 12:36 PM
really i cant think of many reasons it'd be a good idea, but (running in the face of meme's suggestion of people's lack of interest in interactivity) I guess there could be the capacity to edit and reorganise music. thinking that, much in the same way that bitmaps can be encapsulated within illustrator documents (one image can be chopped up and reused in a hundred different ways) you could attach a few "real" sounds to your "midi" file like some real recorded drum beats and have them tiled up by the files programming. :D

It's a growing area - there is technology out there that allows you to load songs onto a CD in such a way that each track can be separated by the listener / consumer and remixed... effectively giving the listener access to the multitrack master tapes. In fact, many labels are currently floating the idea around their artists asking if they would object to providing the multitrack parts of their songs as well as their finished stereo masters.

The most high-profile use of this system so far has been the Prodigy's latest album:

http://www.trust-media.net/concept

But, let's be frank, it's a gimmick. I'd be very surprised if the majority of artists would be comfortable with the idea of letting the listener decide on the final mix of their songs. It's like going to an art gallery and finding that all of the painitings are just outlines and visitors are issued with a pack of Crayolas at the door.

I'm a great believer in the "magic" of making a record - the journey from inspiration to studio to recording to mastering to packaged product. I love the controlled chaos of each link in the chain - the fact that a snare sound on a CD is so far removed from the physical object, yet remains immediately recognisable as a snare drum....

drum shell
skin
stick
player
microphone(s)
room acoustics
preamp
EQ
mixer
storage medium (tape, ProTools etc)
processing
mastering

...they all affect the sound. Add to that list the more recent tools of music-making that can pick up that snare sound and recycle it:

sampler
filters
plug-ins

...and you see that (admittedly in an abstract, romantic sense) there's more to this music lark than just sound waves. I'd rather keep it this way. Leave the final mix to those who had the motivation, inspiration and abilities to make it happen in the first place.

machine hugger
18-10-2004, 07:14 AM
interactivity with music is boring.

machine hugger
18-10-2004, 07:19 AM
I forgot to mention that Hint is my new friend.

Rambler
18-10-2004, 09:18 AM
To cross post back to what Mwanji and Matt were saying...

The old triad system of music has worked pretty well for a long time, and it's not something I object to of course, but it seems to me that something like MIDI is - for all its strengths - a limited application of music technology. Other configurations between composer, performer, listener and score (musical artefact) are possible - jazz is one obvious example where the relationship between composer, score and performer is radically different from that in most classical music; Earle Brown's graphic compositions are another. There are ways (and jazz and Brown, in some ways represent this) of transforming the idea and role of a score. When you write using MIDI, you are using a fairly direct analogy with writing on paper (in fact the best score writing programmes are often considered to be the ones that come closest to the 'feel' of writing on paper), but with that comes the authority of the score. I think the alternatives - using a score as a toolkit for improvisation, performer-based creativity. This could even be listener based. Not in the ghastly sense of audience participation (which is, granted, very boring) but in a way that the private listener at home could play something, create something from this electronic toolkit, for themselves. This is really just an updating of having a piano in the study to tinkle around on, but that isn't necessarily a bad thing - it was a great source of pleasure for many people for many years.

Timbre and classical music, Matt, is a whole other can of worms which I might return too ...

shaun L
18-10-2004, 12:12 PM
Possibly reiterating other peoples ideas...
For midi to replace audio as the future of music distribution, it would have to be able to describe the timbre of the sounds being played... the midi score would have to contain a mass of information relating to harmonics, formants... the wavestructure of the sound.. plus all the products of an instrument's interaction with its environment and its operator.
So... i think that midi has a similar relationship to music as a map does to terrain... a map that was larger/ more bulky/ complex than the terrain it described would not be particularly useful. Audio files may be the simplest way to encode complex sonic data.

General Midi as the future of music!!!!

MolexRoots
18-10-2004, 04:25 PM
I've had almost minimal experience with MIDI in a professional environment. CreativeLab's SoundFont technology was good fun, replacing the soundcard's onboard GeneralMIDI wavetable with a custom made bank. Final Fantasy VII used MIDI throughout the game (on PC) so I built my own SoundFont with some nice quality samples and made the music sound awesome without messing up Nobuo Uematsu's compositions. Trouble is, those banks can get very large. The one I was using with FFVII was about 200MB. And you gotta load that into RAM! Performance hit or what...

mms
19-10-2004, 10:34 AM
some midi, abba and sabbath midi
http://www.sf2midi.com/

Rambler
19-10-2004, 10:43 AM
Which reminds me - how about some hilarious Metallica MIDI?:

Master of Puppets (http://209.197.86.65/20041007/rock/metallica/MasterOfPuppets.mid)

MolexRoots
19-10-2004, 11:58 AM
Nice link, mms.

Rambler
25-10-2004, 03:44 PM
For a composer's-eye view of working with MIDI, you could do worse than click here:
http://www.artsjournal.com/postclassic/archives20041001.shtml#89873

Eppy
26-10-2004, 03:07 PM
My first experiences with writing electronic music were actually in the form of MIDI--specifically, with Cakewalk. I didn't have the money to buy a keyboard or even a module, but I could download Cakewalk and make stuff that way. It was fun, but as my freshman year roommate put it, it tended to sound a bit too much like "video game music." The quality and range really opened up when I started using FruityLoops.

On the other hand, I don't compose string quartets anymore, more's the pity.

I used trackers for a while, but although they were fun to do live stuff with (a lot easier to drop things in and out of the mix and fuck stuff around than other stuff that was extant at the time, at least for me), they were pretty limiting in terms of what they could do with samples. I mean, it was kind of badass to be composing in DOS, but still.

On the other hand, I thought FruityLoops was kind of limiting too until Tom Ellard said he did all his work in it--and gave some pointers on how to really fuck around with the thing. You can find them in the research section (http://www.sevcom.com/ResearchFrameset.html) on Sevcom (http://www.sevcom.com).


On another angle, there's a long history of music technologists trying to create formats that allow music to be released as an interactive medium. When quadrophonic and then multichannel audio formats were developed there was talk of releasing albums in multitrack format to allow users to do their own mix. Same with multimedia -- one of my CEOs used to go on about multimedia meaning listeners could "conduct" classical music.

This failed for two reasons. One was copyright problems. The other was that a lot of the value of music is in the mix and the mastering -- basically, most people can't be bothered to do it, they like having the finished article. The AIFF-quality files required would probably be a bit big for 512K broadband.

This is too bad, as I've been interested in seeing stuff like this for a long time. But I think it's a lot more possible now, if the interest was there. You could stream the tracks into a Flash app that had a virtual mixer to go around the copyright issues, and you could have some sections pre-mixed so you weren't getting all the channels (the drums would all be together, the backing vocals, the guitars, etc.). You wouldn't have total freedom, but you would be able to change it a lot in terms of EQ and things like that. I'd love to see it. Plus, with the new era of creative commons copyrights and open source, I think musicians would be interested in doing a more transparent version of this as well, at least to the degree that they're unencumbered by labels. I wouldn't want unmixed files to be the ONLY delivery option, but to have it there as a possibiltiy would be nice.