My biggest theme this summer was the " completion" of Schuberts Unfinished symphony on the basis of his sketches!! Here the Scherzo with an orchestration of the sketched main part and a new composition of the Trio with Schubert's sketched melody:
https://youtu.be/FSndEjJPeTM?si=62TpTBfnH2xbq4
And here the Finale :
https://youtu.be/L3yOwZ8OEns?si=pBMOy5FklZ4tFCbn (read the remarks in the description of the video!)
What do you think? Of course I would agree when you say that actually the work is already "complete" with only the known 2 previous movements ...
Gerd
Replies
Still, it's an impressive undertaking on your part.
I am an AI skeptic. Being a programmer, I did some research into the algorithms underlying the current AI hype, and discovered that it amounts to nothing other than a large-scale interpolator. I.e., given n data points, what we call AI simply interpolates between them. IOW it amounts to nothing more than just drawing lines between dots, a child's exercise with no insight whatsoever. The only reason it's so "successful" today is because the internet gave access to huge mounts of data -- the more points you have, the more accurate the lines between the dots become -- obviously. But that does not change the nature of the algorithm.
Like pixels on a computer screen, given enough pixel density you can represent a photograph to arbitrary accuracy. However, that says nothing about what the photograph represents. Just because you can produce a close-enough approximation to something, does not mean you've actually reproduced the thing itself, or understood its nature. All you did was to add more pixels to the screen; without the underlying data your AI is still as dumb as rocks. It can only interpolate between data that's already there; it cannot *create* new data that's outside of the dataset it was given.
OK -- the return of the main theme of the symphony. This is not in Mackerras/Newbould and I think it's eminently suitable -- his version seems perfunctory in comparison. As the preceding music actually written by Schubert is pretty unsubtle and bombastic, I'll let what you've done pass. However I'd ditch bar 413 -- the theme doesn't need this preamble in my view.
Just as a general point -- I have little interest in the rights and wrongs of the countless completions of this symphony. I'm not a Schubert scholar or academic so don't care too much about authenticity. But actually I think your version, despite a few question marks, has more character than the comparable Mackerras so well done indeed on that.