Prepare yourself for a full palm to face application, because I didn’t, and the bruise, it hurts. Here is a recent Popular Science article describing a “New Software Processor [that] Can Transcribe Music From Any Performance.”
I am going to describe my first few moments with this article. I see the title and think, “what is a Software Processor?” I guess the software, “it does processes,” like LOL cats does speek. As with so many science popularizations, there are words here with precise meanings but put together in strange ways.
Then I get to “Can Transcribe Music from Any Performance.” That is revolutionary since this “software processor” can transcribe a performance in the most reverberant of places, and even in outer space. At this moment my skepticism bells are tolling like Montmartre on a dimanche morning. How come I haven’t heard of this revolutionary work? Is this going to be another plug for Melodyne?
Then I reach the by-line: “If only Mozart had this.” With this I know I am exactly in the placidly naive waters of the media, where emphasis is put on overreaching claims, and absurd scenarios. What an insult to Mozart. Had he to wait for the computer to boot up every time, the world would be at least 41 symphonies and a Mass short: “Hey Tony! How do I turn off the swing quantization? And how can I see the rest icons in this Finale?”
Then almost simultaneously, I reach the photograph of an obviously staged conductor with cellists unnaturally obedient, and smokey purple lights, and my god, green lasers. At this point my initial to present magnitude respect was hovering at -100 dB.
And then I read further to find that the paper that proposes and tests this “new method to generate sheet music based on the sounds of individual notes” was my Paper of the Day (Po’D) on April 15, 2010. That this novel work received such revolutionary mention like this is completely face-palmable. Even in the Popular Science article, it is said, “As of now, it only works for one instrument at a time, but the researchers think the method can be scaled to include many instruments playing at once,” which contradicts its very headline. Furthermore, if Mozart had to wait for dictionary learning and sparse approximation, the world would be short one Mozart, and many of the composers that followed.
Perhaps the funniest part of this entire thing is that the authors of the research article show that their approach is not as accurate as that of P. Leveau, E. Vincent, G. Richard, and L. Daudet, “Instrument-specific harmonic atoms for mid-level music representation,” IEEE Trans. Audio, Speech, Lang. Process., vol. 16, pp. 116-128, Jan. 2008. This clearly means that the approach of Leveau et al. “Can Transcribe Music from Any Performance NEVER Recorded!” I will contact Popular Science immediately. Using Electronic Email. And my software mailer.