Disklavier

For what it's worth, I would like to chime in and add my two cents about 
scanning in music.

Let me begin by saying that I am an adult trying to learn to play my 
DC3A disklavier.  After spending a year with Voyetra's "Teach Me Piano" 
software which I enjoyed very much and felt was an excellent way to 
begin to learn to play the piano (for an adult), I began taking one hour 
per week piano lessons slightly over one year ago.  The core of my 
instruction has been the Suzuki Piano Method and I am now on the 3rd 
selection in book 4 of this method for those familiar with the method. 
 In addition my instructor always assigns me classical pieces from 
several other books including the Hanon and Czerny technical materials. 
 I would thus classify myself as a very inexperienced student attempting 
to learn the piano.

I began scanning in sheet music about two years ago using Smart Score 
and Finale.  Smart Score V. 1.3 (which I began with) did a very poor job 
in terms of scanning accuracy although this improved somewhat when I 
purchased a newer Epson flatbed scanner.  I can definitely state that 
the quality of the scanner plays an important role in scanning music in 
terms of the accuracy of the results.  Another important issue is the 
interface between the scanning program and the notation program.  It is 
always necessary to correct errors, add dynamics, musical expressions, 
text, etc. in the notation program.  Finale and Smart Score work 
together, but in my opinion not very well.  It is necessary to save a 
Smart Score music scan as a .fin file which must then be imported into 
Finale.  The importation into Finale introduces errors that were not 
present in the scan requiring further correction.  Furthermore, in my 
opinion, Finale has a difficult learning curve (I have been using it 
steadily for 2 years and I still find myself referring constantly to the 
manual).  While Smart Score has improved significantly with Version 2.1 
(currently latest version), it still has problems including the fact 
that my computer (running Windows 2000 Pro on a 1.8 GHZ Pentium 4 with 1 
GB of memory) crashes after every single time that I use Smart Score. 
 (No other software that I have produces these crashes but Smart Score.)  

Recently (about two months ago) I purchased Sibelius 2.1 and its 
associated scanning program (Photoscore Pro).  These two programs are 
tightly integrated although Photoscore is more limited in its 
recognition of musical symbols than Smart Score.  Both Sibelius 2.1 and 
Photoscore Pro 2.1 are very very stable.  I have never experienced a 
crash with either program even when running other programs and using 
them for hours.  Photoscore's note accuracy in music scanning is 
excellent with my Epson flatbed scanner.  With all of that said, to get 
a piece to play with correct interpretations of the dynamics and other 
expressions in the score requires extensive editing in Sibelius. 
 Fortunately Sibelius is much more intuitive to use than Finale, and in 
only 2 months of working with it I feel far more competent to edit 
notation in Sibelius than in Finale.  Further, Sibelius has some very 
interesting artificial intelligence algorithms which can, to some 
extent, interpret correct musical phrasing and apply rubato in 
controlled amounts to the midi playback of the piece.  I find that the 
playback of Sibelius files on my computer using an Creative Live 
Platinum sound card is very good, and the playback of dynamics and 
pedaling on my Yamaha disklavier DC3A is also very good.  

No one should believe that music scanning can produce a file that plays 
back with all of the expressional nuances that an excellent pianist can 
incorporate into a piece.  We are years away from that point, and in 
fact, I would seriously doubt that computers will every seriously 
challenge a good pianist in terms of playback.  With that said, I do 
find the playback of the music that I have scanned in using Photoscore 
and edited in Sibelius and Finale to be musically satisfying.  That 
statement may reveal more about my lack of musical sophistication than 
anything else, but I have also had these files played back for others 
(including pianists) who are quite amazed and pleased with the quality 
of the playback.  

The bottom line is that scanning in of  polyphonic music (which nearly 
all piano music is) is not a trivial or inexpensive process. It requires 
a great deal of patience to learn these programs, understand what they 
can and cannot do, and a willingness to spend many hours to build the 
kind of playback that will satisfy most,  but not the most 
discriminating listeners. However, it can be done, and the results of 
music scanning have been most helpful to me in learning new pieces, and 
in listening to a wide assortment of music on the web for Finale and 
Sibelius.  

Tom