 |
|
You are not logged in! F.A.Q Log in Register |
 |
|
 |
 Now online (1)dariusgriffin ...and 571 guests Last 5 registered Oplandisks nothingstar N_loop yipe foxtrotromeo Browse members... |
 |
Members 8025Messages 2614124 Today 3 Topics 127542 |
|
|||||||||
| |||||||||
|
|
hello all, where are the afx midi files gone!! the links on xltronic return 404s and the search engine isnt working, so if someone could help me that would be wicked. any / all afx midi files, not just drukqs ones Im doing a project building a program which analyses patterns in midi files and then generates new files based on what it has seen ; essentially writing new pieces in the style of a composer. cheers |
||||||||
|
|
|||||||||
| |||||||||
|
|
sounds very interesting can u post some more details about yr program what does it use to analyse it ? how does it work ? have u used symbollic composer? |
||||||||
|
|
|||||||||
| |||||||||
|
|
oh sorry dont know whwere the midifiles are btw |
||||||||
|
|
|||||||||
| |||||||||
|
|
its for my final year at university. i'll be writing the program in Java, it will (hopefully!) read and analyses a load of midi files that i feed into it, so it can see which notes commonly follow other notes. This way it will build a stochastic matrix which will be used during the genertaion process using markov chains (i.e. it will see that its just generated a note c1 then e1, then will look up to see what the next note should be). pattern recongnition and discovery is anohter problem which is harder to solve, fair enough you know which notes go well together now but how do you get it to structure those notes in a creative yet sensible way? ive only been researching it for 2 weeks so im not sure which algorithms i will use. my supervisor should be able to help me though; hes been researching this shit for like 20 years. |
||||||||
|
|
|||||||||
| |||||||||
|
|
4 is quite an easy one to find. |
||||||||
|
|
|||||||||
| |||||||||
|
|
cool! ive been meaning to read up on markov chains for ages they crop up a lot in algorithmic composition stuff it seems |
||||||||
|
|
|||||||||
| |||||||||
|
|
eeeehhhhh.. one large flaw: if you indiscriminately input random songs, and by coincidence, most of those songs are in for instance E minor, you'll get a "biased" result. I understand that you're just doing the program, but could you please try to input an equal amount of songs with each scale (of the scales you encounter/input) so that the result of the calculation ultimately may be less biased and thus in some weird way possibly valuable or even more general? |
||||||||
|
|
|||||||||
| |||||||||
|
|
either diversity or confinement... if you confine yourself to one scale, you could produce "the ultimate song in e minor" or something... |
||||||||
|
|
|||||||||
| |||||||||
|
|
i took a class on exactly this kind of programming last year called 'a.i. and music.' it was interesting but didnt give me anything to work with in my own music, mainly because all of what we learned is in this archaic language 'lisp.' which just extremely basic. ps ive only seen a midi for 4, and i dont know where it is. g'luck. |
||||||||
|
|
|||||||||
| |||||||||
|
|
|
||||||||
|
|
|||||||||
| |||||||||
|
|
nice to see so many people interested in my little project! steve mqueen: markv chains are very interesting website explains it quite well and gives some java code too drunken masta: yes that is a problem i need to try and avoid if im generating new pieces in a new style. however, if i just feed 20 satie pieces into it, so i want new pieces to sound like satie, maybe the majority or of songs start off in a particular key and the program will pick this out for the new one. but yeah for 'general' new pieces i shud put in a good distribution of files hed: yeah LISP looks pretty boring, im not touching it |
||||||||
|
|
|||||||||
| |||||||||
|
|
please let me know if you find any others. |
||||||||
|
|
|||||||||
| |||||||||
|
|
i want drukqs midi files, that would be entertaining! :D |
||||||||
|
|
|||||||||
| |||||||||
|
|
"of what we learned is in this archaic language 'lisp.' which just extremely basic. " You dont know a fucking thing about LISP obviously. its beautiful. best language ever invented imo |
||||||||
|
|
|||||||||
| |||||||||
|
|
thanx 4 the link grinningcat .. thats a pretty short markov prog !! i got a lib extension for supercollider to do markov chains somewhere but it was huge |
||||||||
|
|
|||||||||
| |||||||||
|
|
Would it not be confusing for the program if the midi files had drum programming on them? If you are doing this via key and note selection the drums would throw this off, right? I might be being dumb though, so explain it to me. |
||||||||
|
|
|||||||||
|
Messageboard index
|
|||||||||
