Just read a paper from ismir 2009 that several music snippets are remixed according to their tonal and tempo information. I mean, I have this idea very long time ago... But I did not do anything and now it is others' contribution. Though I am not saying I will be able to solve the question, I am just lazy and think way much more than work.
Anyway, if the technology is already there, what else can we do? Or can we go beyond this? Therefore the application idea pop out, simply an iphone app. If we can pick two songs and shake the iphone which indicates that blend/mix this two songs together in "a way". Can be if shake harder, they will blend better like phrase by phrase blend or something. And the technology required are tempo estimation as well as key detection with phase vocoder to synchronize these two attributes.
2 comments:
Reminds me of the "blender" on real DJ turntable. Basically the DJ play two songs simultaneously and use the blender to switch back and forth between them. Of course the DJ had to match the tempo and key manually.
If you can design a analyzer that will analyze the current song, and not only find a list of songs from the database that matches in tempo and key, but also generate various "cue points" in each song; then you can use shake right/shake left to switch back and forth between songs and various cue points. That will be really cool.
Check out "Pacemaker" http://www.pacemaker.net/ , it does tempo matching automatically, and comes with a blender.
P.S. I always think the "shake" function in iPod and phone is just for flair...
The "blender" I mentioned is called "cross-fader" actually.
Post a Comment