Here is a new musical christmas experiment: I chopped 4 compact disks of children musicals by the German composer and all-round musician Reinhard Lakomy [1] into small pieces and re-assembled them to match a famous christmas song. As chunks lengths I tried both about 1/6s [2] and 1/3s [3]. Generally it holds: The longer the chunks, the more you can recognize the origin of the pieces. The shorter the chunks, the better is the approximation to the christmas song (and the longer takes the computation). If you want to guess the song, then click now and ignore the video title. :-)
[1] http://de.wikipedia.org/wiki/Reinhard_Lakomy [2] https://www.youtube.com/watch?v=aJWp4pkIkQE [3] https://www.youtube.com/watch?v=jxC5bcwXCyQ The first collage is blocked in Germany, since YouTube thinks that it is too close to the recording of the christmas song. However, no wave from the christmas song recording was copied to the collage and honestly, the collage sounds very different from that recording! YouTube believes to recognize the second collage as the Greek version "Touli Gia To Hristouli". It seems to me that YouTube's recognition algorithm exhibits a sense of humor in this case. You can download the project's code from the darcs repository: [4] http://code.haskell.org/~thielema/sound-collage/ -- Read the whole topic here: Haskell Art: http://lurk.org/r/topic/27SkS6ZdMoRhcQnOlC44EC To leave Haskell Art, email haskell-...@group.lurk.org with the following email subject: unsubscribe