I simply don’t understand the concept of people advertising a product as if it is commercially ready, yet when presented with the most rudimentary tests, it consistently fails in ways that would be evident from even the most basic testing.
Sure, I understand there can be some difficulty in recognizing the difference between slurs and ties and other things like that. But these programs all fail on more basic, obvious things. You can’t get through any moderately difficult pieces of 2 or more pages without errors that should have been fixed at the alpha level of the product.
I really think these companies are badly missing a business opportunity. I think there would be considerable demand for a product that doesn’t make so many basic mistakes. Even with the considerable improvements these two new versions demonstrate, it still seems questionable to me whether there is a net time savings. I have used SmartScore in a few limited cases in the past with some time savings achieved. But there are so many errors that I only have used it on selective projects that would have taken an enormous amount of time to enter from scratch. I can live with not all the dynamics coming through as intended. I cannot live with it failing to recognize key changes mid-staff. I cannot live with it turning multi-measure rests into garbage, throwing the measure counts off wildly.
Here is an example on the very first measure is a very straightforward file with a very legible scan. The people offering products in this space obviously have convinced themselves that 98% is good enough or something like that. Imaging a spreadsheet that calculates formulas right 98% of the time. I don’t expect perfection, but this kind of error is unacceptable to me. Notice that the triplets were recognized in measures 3 and 4. My guess is that the text recognition threw off the music recognition in the first measure. The text recognition is pretty much worthless to me, and I don’t mind entering tempos and other markings, but the music needs to be right. BOTH products failed in that triplet in the first measure.
As a practical matter I will probably upgrade my SmartScore license because the new version is definitely better than X2. But no way I’d pay $250 for a new license for programs that are so problematic. If Neuratron had a crossgrade price, then I might buy that instead because it appears PhotoScore may be slightly better than SmartScore. But they don’t, which seems like a dumb move to me.
I have the recent version of Photoscore Ultimate, and it fails always.
I also tested SmartScore before purchasing Photoscore Ultimate, and it was same.
I really would like to compare the results of recognising a scanned orchestral score produced by some users here.
what is the source material you are using?
I get acceptable results. The main key seems to be the quality of the used scan.
Just updated to PhotoScore Ultimate today. Definitely more comprehensive in its recognition than the Pro version I’ve been using. Not too happy that it seems to be focused around Sibelius (ie. ‘send to Sibelius’). Having said that it does it’s job reasonably well but it certainly doesn’t have a Mac looking interface (another downer) and looks quite old fashioned. A plug in for Dorico would certainly be welcomed although I suspect may never happen considering Avid’s Espressivo technology is embedded within it. It could certainly do with a facelift though.
Would make more sense, to write this to Neuratron, the company who owns and develops PhotoScore.
I agree with HeiPet, especially because I already wrote them to basically ask for a more Doricoish workflow.
Not sure they really care, though. But the more we write, the better, I suppose!
Edit: I have found that Photoscore ultimate is really more useful in its version 9 (paid update, alas). But placing slurs and dynamics correctly is still compulsory (and I still do not recognize any text, as this is really a waste of time).
It is some time since I abandoned Photoscore because of the time and effort needed to make multiple corrections. I was using it to import pdfs of good engravings of full scores of the symphonies of Beethoven and others, and I found it more reliable to input these entirely by hand. So it is disappointing to read here that little progress appears to have been made since then.
Have written to Neuratron and my comments have been passed on to the development team.
Very good. A “development team”? I always had the impression, that it is more or less a one man show of Martin Dawe …