Here are some quick observations from a first test of the Smatscore 64 demo:
- It did not start well. I tried to drag a PDF into the open desktop app. It apparently isn’t supporting drag and drop. That’s rather disappointing considering how much they talked about workflow. I was able to open the PDF going through File Open. But even then, it seemed like the program didn’t respond to some actions. I eventually got it to interpret the PDF (which was a raster scan).
- The recognition completed without any exceptions and looked pretty clean.
- This was a well engraved music source with a clean scan. But it was a piece with numerous elements that have been challenging to all scanning programs.
- The quick assessment is that Smartscore 64 definitely scans better than X2 Pro. On this particular piece, With the exception of 3 measures in 3 pages, I didn’t miss any notes or accidentals. Almost all the key signatures and clef changes (a big problem in the past) came through fine. And amazingly, one of those 3 measures was a misprint in the original, with a measure having a dotted eighth rest that should not have been dotted. So SmartScore recognized the music and also flagged the incorrect number of beats.
- It is still getting too much wrong. The very first measure missed a quarter-note triplet that is perfectly clear. Most of the slurs seem to connect to the wrong notes. With the demo, I can’t save the XML to see how it imports into Dorico, but I fear those slurs will be a hot mess. It got one of the multi-measure rests beautifully and made a complete mess of the other. There was a sequence mid-staff changed from 4/4 to 2/4 for a half rest, followed by a double bar, key change from C to Ab , then 4/4 meter. In this sequence the key change was completely missed. That’s the kind of thing that makes these scanning programs so marginal. Many of the triplet brackets caused the first note of the triplet to incorrectly get a staccato.
- It got most of the dynamics – probably 90% of them.
- The playback was an octave too low.
So I am of mixed feelings. It is definitely an improvement, and maybe now solidly in the area of doing more good than harm. But it has been 6 years since the last upgrade. I fear I will not live long enough to see one of these programs actually work. Given how many bugs I found in 15 minutes with a very clean score, it appears that nobody really cared to test this program very much.