Home > Culture, Misc, Music Discussion, Technology > Idea / Invention: Music recordings that sound different on each play

Idea / Invention: Music recordings that sound different on each play

June 11th, 2010

I had a neat idea as I was driving to a gig. I get really fed up with the recordings (stack of cd’s) in my car as I’ve listened to them so many times, and every time I listen to them, they are the same!

I thought, why not make a recording that sounds different every time you play it?

Sound crazy? Well, it’s really quite possible to do, and wouldn’t be very hard.

Usually when a band records a song, they might do multiple takes, say 4 or 5 takes. For certain sections of the song, such as an instrumental solo, it would be cool if it would sound different every time. The artist or recording engineer could “program” the song to play a different, random take for a certain section of the song, each time the recording is played, or combine together parts of any of the takes in a new or somewhat random way. A single solo section could be further broken down into segments, such as a certain number of bars, and each segment could be swapped in with a different take, as the music is played.

There are a bunch of extensions to this idea:
– if this is on a website, allow the user to save their particular performance that they heard, and share it with others (and rate them, etc)

– instead of a purely random choice of the takes, the user could give input into the choices such as “I’d like a really upbeat and lively version of this song”, or “I’d really like a version of this song with a longer sax solo and less Britney”.

-if the “intensity” of the performance was hooked up to a foot pedal (e.g. the recording would become more excited as the pedal is depressed lower, and more relaxed as the pedal is released) an actual live musician could play along with such a recording, while controlling the intensity of the background recording, and have the background music follow his/her performance.

Please – let me know what you think!! And please take this idea, expand on it, and develop something new, as I mostly have time to think these days and not implement something, even so cool as this!

Geoff

Update, Sept. 11, 2010: I’ve created a player which implements this concept, and made it open source. Check it out at MultipathAudio.com. Includes an example song.

4,019 views
  1. Angelica
    June 12th, 2010 at 11:18 | #1

    Yeah! Great ideas. One of the members of my lab is doing some research that could make this possible, using synthesis though:
    winnie.kuis.kyoto-u.ac.jp/members/yasuraok/

    I like the idea of using real recordings, and the applications you mention. I’ll pass them on 🙂

    Have you heard of UJAM? www.youtube.com/watch?v=HxURUkfI7yY

    It takes a melody and transforms it into different styles with full orchestration. Maybe this could be hooked up to accept existing, full recordings 🙂

  2. Geoff
    June 12th, 2010 at 11:18 | #2

    @Angelica
    thanks Angelica!! I just checked out UJAM, it looks interesting.

  3. Colin
    June 12th, 2010 at 11:20 | #3

    Yes! Sounds like more work for mixing engineers / composers, but it would be worth it for the easily musically bored, like us! It would work well for heavily produced music that is click-tracked and multi-layered to start with. Then you could set up a bunch of “permissable combinations” of sounds/mixing levels/takes/effects. It would be like a bunch of dub mixes of the original track.

  4. Geoff
    June 12th, 2010 at 11:22 | #4

    hey Colin, thanks for the comments! 🙂 hopefully when I have some free time I’ll make a little “proof of concept” web app…

  5. September 1st, 2010 at 18:35 | #5

    No, sorry, I don’t buy it. If you get tired of the CD, just don’t listen to it anymore. I stopped listening to music like I used to. I don’t think it should be over-listened. It is like eating good chocolate — you don’t want too much of it.
    I would find very interesting some automatic way to mix existing music, and I have seem something like that done in a company I worked for, called Echonest. It was a computer dj, that was smart enough to piece together different parts of songs, to make it sound like a new song.

    Another interesting idea would be to record not a song, but just the interesting part of a song — a phraze. By phraze I mean something very small that sounds good and unique. So you’d be listenning to a collection of phrazes. Then you could have an automatic software to put the phrazes together in a random way.

  6. Geoff
    September 2nd, 2010 at 11:16 | #6

    Hehe, Boris, thanks for the comments. Sorry you don’t buy it, but I’ve already got a proof of concept working at home, and it is really awesome in my opinion. Stay tuned! I will be releasing a track from my next CD using this “Multi Path Audio” system.

  7. September 11th, 2010 at 16:55 | #7

    Update, Sept. 11, 2010: I’ve created a player which implements this concept, and made it open source. Check it out at MultipathAudio.com. Includes an example song.