Pop Song Writing Patch


    Mar 26 2007 | 12:27 pm
    Im a University student and for my final project i would like to build a patch which writes pop songs...
    Choose a tempo Select a key select an arrangement, (four bar loop, bridge, drum fill in right place, extra instuments in chorus ect..) random drum pattern random drum kit random timres for base, lead, ect Melody with phrases which make sense (not completely random) Ect... ect..
    And i would like it to sequence it all first (visually) before it plays.
    I think you all get the idea. Has anything like this already been done and published? Does anyone have a link they could help me out with.
    Thanking you in advance
    Boskins

    • Mar 26 2007 | 12:55 pm
      Quote: Boskins wrote on Mon, 26 March 2007 06:27 ---------------------------------------------------- > Im a University student and for my final project i would like to build a patch which writes pop songs... > > I think you all get the idea. Has anything like this already been done and published? Does anyone have a link they could help me out with.
      It's not supposed to be your own work, then?
    • Mar 26 2007 | 1:59 pm
      well kinda, weve been told we can get bits from other places as long as we can explain how it all works in our write up.
      thanks
      So any suggestions
    • Mar 26 2007 | 2:02 pm
      I'd say what you've listed is very possible in Max/msp. It is a LOT of work though, and you will need to work well structured to obtain a decent final product.
      There are many sequencers built in Max, but are you allowed to copy others' work at all? If you want to build the whole thing from scratch, i'd say you need at least 3-5 months, provided:
      - You know Max/msp and have experience working with msp. - Your plan is well designed and doesn't need any revisiting. - You don't build parts that you replace later on. - Your UI doesn't need to be top-notch.
      This is just a rough estimate of course, keep in mind that i don't know how much of a genius you are (jkc could probably build all of it in about a day, haha).
      Are you planning to make the patterns yourself? Are these just a lot of XML/TXT patterns from which you choose (parts) at random or should the actual thing be 'intelligent', i.e. generate melodies purely from keys and a couple abstract params? Will you be using the same instruments all the time?
      I'd seriously question whether the scope of this project isn't too large. You'll spend a lot of time structuring your work and running into new problems. Also, what course do you do? Computer science, AI? On the other hand, final results don't have to be of publishable quality, so if you can defend your work and your choices you might be ok. It's a risk though. That's my 0.02 dollars.
    • Mar 26 2007 | 2:09 pm
    • Mar 26 2007 | 4:32 pm
      thanks for the responses guys...
      Thats so dissapointing! Ive litterally just started using max/msp and got all excited at the pospect of using a programme which litterally had unlimited possibilities. Ive only got about 90 days to build something... so it looks like i cant do it.
      In answer to someones question, yes i was hoping it would be intelligent and create its own melodies whilst being restricted by a few parameters. I was going to use it to make mainly 'hip hop beats' and due to their simplicity thought it would be a realtively straight forward project. How wrong i was!
      Would anyone be interested in building such a patch?....so that i could buy a copy off them (nothing to do with university).
      Thanks again guys
    • Mar 26 2007 | 11:19 pm
      Alas you may be on the wrong track of questioning - its not what you ask - but how you ask it- people get bugged if you ask for everything on a plate, but if you try and make it work and post the patch that doesnt work well.Alot of the time someone may come up with something to help!. Your best bet thus is to learn some bits and bobs for yourself - and find a way to make it work.
      Anyway, rambling aside, people are probably not going to do the work for you but, there is patches out there that do similar to what you want (though i cant think of them currently) For help getting the general priciples - Have a look at snot Wong's stuff http://www.geocities.com/snottywong_1999/maxmsp/index.html - and see if anything there'll get you thinking in the right direction. Other than that do the tutorials, in a few months you'll be able to make sense of max enough to realise whether you can make it work or not!
      Good luck
      T
    • Mar 26 2007 | 11:34 pm
      I wouldn't let this commentary discourage you necessarily.
      I think one of the most important and useful things to keep in mind when building any type of software is that you start with simple modules and add features as you go. Don't start out trying to develop the comprehensive pop song writing patch. Start with a patch that plays and selects drum sequences. Add on key shifting.
      If you are using general midi, a lot of the sound design work will be done for you. Just select midi values. Start there at least, and then add synthesis capabilities if you see fit later.
      I think this project is totally doable in your time span, at least in some simple form, and you will infinitely learn a lot about max out of it (which is the point, right?).
      And remember, when designing software, you always throw the first 2 versions away. ;)
      best of luck.
      barry
      John Smith wrote: > thanks for the responses guys... > > Thats so dissapointing! Ive litterally just started using max/msp and got all excited at the pospect of using a programme which litterally had unlimited possibilities. Ive only got about 90 days to build something... so it looks like i cant do it. > > In answer to someones question, yes i was hoping it would be intelligent and create its own melodies whilst being restricted by a few parameters. I was going to use it to make mainly 'hip hop beats' and due to their simplicity thought it would be a realtively straight forward project. How wrong i was! > > Would anyone be interested in building such a patch?....so that i could buy a copy off them (nothing to do with university). > > Thanks again guys >
      -- barry threw composition : sound : programming http://www.barrythrew.com bthrew(at)gmail(dot)com 857-544-3967
      Today, Noise is triumphant and reigns sovereign over the sensibility of men. - Luigi Russolo, The Art of Noises
    • Mar 27 2007 | 4:56 am
      damn, this patch sounds difficult. I wonder if he'll wuss out ;)
    • Mar 27 2007 | 4:57 am
      I think a cool idea would be one that read midi files of pop songs and created themes and variations and modulations and such.... that sounds even harder. oh my poor, small brain.
    • Mar 27 2007 | 5:32 am
      That was something the original RCA Mark I synthesizer was supposed to do in the early 1950s.
      On Mar 26, 2007, at 9:57 PM, jamez wrote:
      > I think a cool idea would be one that read midi files of pop songs > and created themes and variations and modulations and such.... that > sounds even harder.
    • Mar 27 2007 | 5:42 am
      arne: ^_^ looks awesome!!
    • Mar 27 2007 | 9:46 am
      thanks for all the responses guy. Certainly food for thought!
    • Mar 27 2007 | 12:05 pm
      Quoting jamez :
      > I think a cool idea would be one that read midi files of pop > songs and created themes and variations and modulations and > such.... that sounds even harder. oh my poor, small brain.
      One of our former students, Michael Prerau, did this with a GA a few years back, here's the abstract of a paper he presented:
      and a little demo of the implementation:
      As far as "pop song" stuff, we did some basic blues-algorithms in class, but it was more aimed at modeling performance instead of constructing new songs:
    • Mar 27 2007 | 12:07 pm
      Quoting Arne Eigenfeldt :
      > That was something the original RCA Mark I synthesizer was > supposed > to do in the early 1950s. > > http://www.obsolete.com/120_years/machines/rca/
      Ha ha -- the things you say to get grant money! :-) Our Mark II was supposedly designed to "replace the symphony orchestra..."
    • Mar 27 2007 | 10:23 pm
      John Smith schrieb: > I was going to use it to make mainly 'hip hop beats' > and due to their simplicity thought it would be a realtively straight > forward project. How wrong i was!
      Ah, you should have told that with the Hip Hop, Pop songs are simple but significantly more complex than Hip Hop. I'd say you can do it in 90 days, 3 days for the tutorials, don't use MSP, just Midi. Get a copy of the RTC lib and just start...
      > Would anyone be interested in building such a patch?....so that i > could buy a copy off them (nothing to do with university).
      I am suffering for 20 years now, to pay that would cost you the average income for 20 years... ;-)
      Stefan
      -- Stefan Tiedje------------x------- --_____-----------|-------------- --(_|_ ----|-----|-----()------- -- _|_)----|-----()-------------- ----------()--------www.ccmix.com
    • Mar 28 2007 | 8:11 am
      Thanks Stefan,
      Whats RTC Lib?
      Thanks
    • Mar 28 2007 | 8:30 am
      Never use a questionmark before using Google!
      Its a Real Time Composition Library.
      I think you can make some nice software in 90 days, but the music it will create won't beat the quality of modern ringtones. :)
    • Mar 28 2007 | 4:07 pm
      i made something like this a few years back (it was supposed to be for an installation in vegas but the budget fell through), took about a month, and it came out great. Sorry to be selfish but I really don't want to give out the code. I will tell you that its extremely simple. Pick ammount of chords in progression, use drunk to create a list for melody, bass and then program about 100 different main drum patterns, with random fills. Should take about a month if your half way decent with max, and I also reccomend using external midi instrumentatin rather then programming msp synths to handle the actual sound generation. the internal midi works fine for demo but I would get a motif if your gonna be serious about it.
      I have actually created tons of publishable work this way. Generative is the way of the future.
      NCRIII
    • Mar 30 2007 | 11:16 am
      I think he was kidding.
      If not, it's still a funny statement.
      On Mar 29, 2007, at 4:14 PM, Stefan Tiedje wrote:
      > f.e schrieb: >> You can't. Pop songs cannot be written by computers. You have to >> be gifted, and to buy yourself a guitar. > > This cries for being proved wrong, and mind you a bad pop song is > still a pop song... > And if a computer can write bad Bach cantatas, a pop song should be > a snap... > > And there are so many Pop songs written by ungifted without a guitar. > I always thought guitars are for Rock music only... > And there are some gifted musicians which do it with a piano... > > Stefan > > -- > Stefan Tiedje------------x------- > --_____-----------|-------------- > --(_|_ ----|-----|-----()------- > -- _|_)----|-----()-------------- > ----------()--------www.ccmix.com >
    • Apr 23 2007 | 6:00 pm
      Go for it Boskins. I reckon it's possible. &even if you only get some parts working, you'll get a good idea of what max is capable of.
      For starters, you could just use the Billie Jean beat for the drums. Then concentrate on changing the sounds and picking chords. Chords are of course related to each other. Check out the circle of fifths - you may want not such suprising transitions.
      I had to do a real-time drum tracker last year and really got into it. One thing that helped is Javascript (and later I switched to Java as it's better) You can literally code whatever you want the object to do.
      I know this is a late reply, but hope it helps.
    • Apr 23 2007 | 8:32 pm
      Tried to find the details of this on line. Perhaps someone else recalls a nun at St. Louis in the 1940's or 50's. The name I recall is "Mother Padberg." She was using a computer to compose songs. The news media took note with a headline like "Electronic brain composes 1000 pop songs an hour." If the thing is still running, it is approaching a half billion of those pop songs.
      On 4/23/07 2:00 PM, "Andrew Robertson" wrote:
      > > > Go for it Boskins. I reckon it's possible. &even if you only get some parts > working, you'll get a good idea of what max is capable of. > > For starters, you could just use the Billie Jean beat for the drums. Then > concentrate on changing the sounds and picking chords. Chords are of course > related to each other. Check out the circle of fifths - you may want not such > suprising transitions. > > I had to do a real-time drum tracker last year and really got into it. One > thing that helped is Javascript (and later I switched to Java as it's better) > You can literally code whatever you want the object to do. > > I know this is a late reply, but hope it helps.
      Cheers Gary Lee Nelson Oberlin College www.timara.oberlin.edu/GaryLeeNelson
    • Apr 24 2007 | 1:22 pm
      Gary Lee Nelson schrieb: > Tried to find the details of this on line. Perhaps someone else recalls a > nun at St. Louis in the 1940's or 50's. The name I recall is "Mother > Padberg." She was using a computer to compose songs. The news media took > note with a headline like "Electronic brain composes 1000 pop songs an > hour." If the thing is still running, it is approaching a half billion of > those pop songs.
      I guess Brad is the expert for this, as the only computer in the 50's sort of aimed at that, was at Columbia. There are some mp3's of music done with it. If you listen to that, it's no wonder none of them made it into the top billon... ;-)
      Stefan
      -- Stefan Tiedje------------x------- --_____-----------|-------------- --(_|_ ----|-----|-----()------- -- _|_)----|-----()-------------- ----------()--------www.ccmix.com
    • Apr 24 2007 | 1:57 pm
      The last known recording of the Columbia-Princeton machine appears on the second recording by the Freight Elevator Quartet ("The Freight Elevator Quartet's Jungle Album") from the late 90s. I think it's maybe a minute and a half long. While there are several recorded examples of Raymond Scott's machine, the Columbia Princeton box is much harder to locate if you're looking to listen.
    • Apr 24 2007 | 2:03 pm
      Quoting Stefan Tiedje :
      > I guess Brad is the expert for this, as the only computer in the > 50's > sort of aimed at that, was at Columbia. There are some mp3's of > music > done with it. If you listen to that, it's no wonder none of them > made it > into the top billon... ;-)
      Here ya go:
    • Apr 24 2007 | 2:06 pm
      Come to NIME07 in NYC this June!
      We're hosting the conference activities on the last day (June 10) at the CMC, and at the end of the concert the RCA will SOUND AGAIN! :-)
      (or so we hope...)
      Quoting Gregory Taylor :
      > > The last known recording of the Columbia-Princeton > machine appears on the second recording by the Freight > Elevator Quartet ("The Freight Elevator Quartet's > Jungle Album") from the late 90s. I think it's maybe > a minute and a half long. While there are several > recorded examples of Raymond Scott's machine, the > Columbia Princeton box is much harder to locate if > you're looking to listen. > -- > knowledge is not enough/science is not enough/Love is dreaming > this equation >
    • Apr 24 2007 | 3:06 pm
      I find the Stephen Foster Medley stunning. This stuff is 50 years old, right? More convincing than the Yamaha CX5 computer music demo from the eighties...
      Best,
      Zip
      Zip Boterbloem Media Mechanics Zwaluwstraat 54 2025 VR Haarlem The Netherlands +31627014758 zip@knoware.nl
      Op 24-apr-2007, om 16:03 heeft garton@columbia.edu het volgende geschreven:
      > Quoting Stefan Tiedje : > >> I guess Brad is the expert for this, as the only computer in the >> 50's >> sort of aimed at that, was at Columbia. There are some mp3's of >> music >> done with it. If you listen to that, it's no wonder none of them >> made it >> into the top billon... ;-) > > Here ya go: > > http://music.columbia.edu/cmc/courses/g6611/spring2007/week1/ > index.html > > > brad > http://music.columbia.edu/~brad