Monday, 20 February 2012

Who will save the forgotten music

I was just listening to Janacek's sinfonietta when my thoughts drifted away from one of my top secret projects and, in my mind, I saw a vision of all the pieces of classical music ever created, one by one sliding into oblivion and dying.  The process is slow but it is happening even now, as I am Swyping these words on my Android phone.  The most well-known masterpieces are still with us, still far from being forgotten, but most of classical music is not being played anymore.  It is all over again the vicious circle of marketing probing customers' demand, selecting musical pieces for which  the measured demand is highest (i.e. the most popular (or, in other words, the least forgotten) music is being selected), and feeding this selection back to the customers.  The music which has not been selected is effectively being slowly erased from our collective memory (and less and less is being selected with each iteration of this cycle as yesterday's less demanded becomes today's least demanded since the previous least demanded have been filtered out from the today's selection list).  How many of you had ever listened to Janacek before Haruki Murakami's 1Q84 appeared?
What could be done to save dying music (and bring all those forgotten pieces back to us)?  I think I have found a solution which would be quite easy to implement (certainly easier than bringing down capitalism with its monetary system and switching humanity to the resource-based economy (which, I hope, will happen someday)).  Hiring a good director and a full orchestra to prepare and then perform even just one piece is incredibly expensive but importing a score into a computer programme and charging a computer with the task of performing such music is very cheap and quick nowadays (I am sure most of us have used a midi player at least once in our life).  Of course settling for just that is nothing new and would not be enough as computers are notorious for their mechanical and dull music performance.  However, if we go one step further and throw machine learning methods into this then we should be able to teach our computers how to give unique and inspired musical performances.  We have all the needed elements.  Machine learning methods have become quite powerful and the data - I.e. recordings of the best and worst performances together with their complete musical scores - is there.  Each performance is different which means not 100% faithful to the score.  It is these fluctuations what makes one performance unforgettable and another just dull.  Finding patterns in these fluctuations and being able to generalise these patterns and apply them to any score is, from my perspective, just another exercise in supervised learning.  Seems like an interesting machine learning project with possible commercial applications (think about economical impact on production of classical music considering the huge savings on live performers and production time).  Too bad there are only 24 hours in a day and I cannot realise all those ideas which appear in my mind every day.  Maybe I should start publishing them in hope that someone will be inspired and see them through ;).  Let us hope the music will be saved.  It would be so nice to be able to browse through and listen to all those now forgotten masterpieces (most of them never recorded) somewhere on the Web.
Apropos inspiration.  I just realised that the chain of events which led to my reading the novel which resulted in my listening to the sinfonietta which in turn made my thoughts wandering and ended up  in my writing this blog post, all that started with my dear friend Yoshi who I should probably thank for this moment of contemplation.  It is good to have friends who can inspire us, without them our life would not be full.

No comments:

Post a Comment

Popular Posts