I have spent a lot of time the past 10 days assessing the progress of the music scanning software. I last used these products about 7 years ago and lost interest because they were just too limited. I couldn't find any cases where it was faster to scan than to just enter from scratch. After diving back into this arena, my first impression was that the progress has been discouragingly slow. Indeed, it seems the only supplier that has had any recent innovation is Musitek with their Smartscore X2 product.
My initial tests convinced me that the error rates were still unacceptably high -- with a majority of measures requiring some additional editing/correction. As I have continued my deep dive, I am coming to a somewhat different opinion, based on two important factors that were not initially obvious to me: 1) While there still are many recognition errors in today's products, the current products do not make many errors on the pitch of notes. And as I look back to my experience 7 years ago, my greatest frustration was that there were so many pitches entered incorrectly that I wasted an enormous amount of time going over the translated score note by note to make sure all the pitches were correct. I am finding I don't have to do that now. There may be rhythm problems, but generally speaking the pitches are reliable unless the original document has problems like slurs passing through accidental marks. And when you combine that with improved playback quality (Garritan etc.), I am simply not spending a lot of time worrying about pitches now. That part is rather clean. 2) While the start-to-finish time inside the scanning program is not radically improved in the current generation, that is not really the important metric. The important metric is how long does it take from start to finish on the project. There have been many little improvements that help the work flow and reduce the time incrementally. Such as: improvements in MusicXM, better recognition of dynamics, articulations, hairpins, etc. And one big step forward for me is that the new Smartscore will use a PDF if it is of sufficient resolution. That means for music that is in letter-sized format, I can drop the whole thing in my scanner's sheet feeder and be ready to start on the recognition within a minute. All those things tilt the balance toward scanning being a viable way to work on some projects. I also have a scanner that can sheet-feed 11x17, but I haven't tried that yet. Based on these improvements,I have upgraded to the SmartScore Pro X2 product. I expect I will be able to use that 3 or 4 times a month at least. I have given up my wish that I could simply push a button and have the music cleanly imported into Finale. That's isn't happening. However for many things I do, I believe Smartscore will save some time. For example, I might take a published piece for solo flute and piano accompaniment, scan it in, and turn it into a trumpet duet with piano accompaniment. Or maybe I'll take all of that into a DAW and add other instruments to make it a richer orchestration. For projects like that, it looks like scanning is viable, as long as one accepts that scanning in 2014 is just a tool to speed up part of the first step. I'd be interested in any other experiences / philosophies. _______________________________________________ Finale mailing list Finale@shsu.edu http://lists.shsu.edu/mailman/listinfo/finale