Scanning in manuscript collection OMR?

[Warrentdo]

Hello All,
Well i thought that I would have a go in turning my many pieces or music from paper to digital.
I'm sure that this has been covered many times and im sure i will be told so but im finding if difficult to find on this site so any pointers will be gratefully appreciated.
I could scan the data into the computer and leave it, but my eyes are not the same as they used to be and I find that there is a difference between the paper and the screen even at 1200dpi.
maybe this should be a two question post?

What are the best scanning options.
I will list my opinions on scanning but I think 600 DPI B+W PDF is working the best.

What is the best OMR software.
have a list of OMR software that I would like to test - Capella, Musiket , Scanscore, Sharpeye etc but has anyone any experience with any or have any opinion.

I will trial and post back my results if you think that they will be useful.
Regards,
Warren.

[voet]

For any music that is public domain, you could go to the International Music Score Library Project and download files from there. That might save you some time.

https://imslp.org/wiki/Main_Page

Good luck on your project.

[Leisesturm]

I just purchased Scanscore as they have just (temporarily) dropped the price on the 'Professional' package to $99. I can't use it yet because it only runs in Windows 10. Today's project is getting my system backed up and ready for Windows 10. I have specific requirements for wanting to use Scanscore for document input. For someone who simply wants a way to archive and have scanned music in digital form I would recommend some kind of iPad and an app for the purpose of scanning and pdf creation. In another thread I gave a link that covers the subject well.

[Warrentdo]

Hello, im also looking at Scanscore, but the Ensemble. it's $7 for Melody and then another $30 to upgrade to Ensemble, so that's $37 in total doing it that way as apposed to $69 buying it.
I don't think I will need professional. Do you know if the only difference is the number of four staves per system? Ensemble is 4 and pro is unlimited?
regards. So far Scanscore is looking good, but it seems to make no difference if scanning at 300dpi or 12300 dpi as SS seems to downgrade it.
also thank you for the other link but i have a 27 inch touch screen monitor on the organ that I intend to use but ive not even looked at music organizers yet, another post im sure. thanks again.
Warren.

[Leisesturm]

Scanscore is intended to be a front end for a robust notation program. As such, it doesn't really care about the resolution of the image beyond what is necessary to get good OMR. It has some interesting notation editing features of its own actually. That is a slick loophole exploit you have found, to upgrade Melody to Ensemble. Hmmm. I still think SS is overkill for your purposes ... well not so much overkill as that its focus sidesteps intrinsics that you may find important. I will be transposing C instrument parts to F for my Horn player, and I will be creating 'guide tracks' for the Praise Band from the .pdf playback feature.

Years ago there was a Windows based program called 'MIDIScan' I think, that was more geared to digital document creation and archiving. I don't know if it still exists. Greater resolution than 300dpi allows you to magnify the image without apparent loss of quality. 600dpi should be plenty to get two 8.5 x 11 pages of music to show on a 27" monitor. That is practically 1:1 magnification. I would think that any good document input program would also come with a music organizing tool but I could be wrong about that.

[Warrentdo]

Hello, im just trying to get SS the best scan to eliminate errors. I'm only using SS to make the manuscript clear and sharp and it may be quicker that tidying up a TIFF file. I still lick the ability to transpose and also play back. I will test SS with the scan files that I have created to see if any work better than others and feed back.

[Warrentdo]

What are the best scanning options.
I will list my opinions on scanning but I think 600 DPI B+W PDF is working the best.

Just an update on the scanning bit
I have scanned in a test piece 'Stormy Weather' using Scanner EPSON WF-7610. Software EPSON scan VER 5.2.2.3 en
I have scanned in using BMP,JPEG,TIFF and PDF using various compression settings and the clearest manuscript was the TIFF.
After lots of scans (over 30) i have settled on 1200 dpi, B&W, Edge fill. TIFF N CCIT group4 615KB (approx A4) i can live with that.
Strangely enough I found that scanning 1200 dpi, B&W, Edge fill. PDF PDF/A standard compression from was only 309kb it became blurry and not monochrome.
scanning 600 dpi, B&W, Edge fill. PDF PDF/A standard compression searchable was only 147kb is was searchable but also blurry.
creating (print) a PDF from a TIFF in gimp using Microsoft printer gave excellent results but inflated the file size to 3093kb.
I think that if I found a good TIFF to PDF creator (searchable) that kept the DPI and kept it 2bit then that would be cool.
I will continue to look around as ideally a sharp searchable PDF will be the best (if I can make one!).
Regards,
Warren.

[myorgan]

I have scanned in using BMP,JPEG,TIFF and PDF using various compression settings and the clearest manuscript was the TIFF.
After lots of scans (over 30) i have settled on 1200 dpi, B&W, Edge fill. TIFF N CCIT group4 615KB (approx A4) i can live with that.
Strangely enough I found that scanning 1200 dpi, B&W, Edge fill. PDF PDF/A standard compression from was only 309kb it became blurry and not monochrome.
scanning 600 dpi, B&W, Edge fill. PDF PDF/A standard compression searchable was only 147kb is was searchable but also blurry.
creating (print) a PDF from a TIFF in gimp using Microsoft printer gave excellent results but inflated the file size to 3093kb.

Warren,

It helps if one knows image file types and how they are stored on the computer.

• BMP and TIFF are uncompressed images, and assign one color per pixel. Therefore, they have extremely large filetypes.
• JPG is a compressed image type and makes up for size by assigning one color to several similar pixels. As a result, these filetypes can use as little as 5-10% of the space an uncompressed file.

I'm not sure what your purpose is in scanning. If it is to do optical character recognition (OCR), then you will want to start with a lossless image type (TIFF).

If you are just looking for a clear image, JPG will work fine because in B&W, the program ostensibly is assigning one of two numbers to multiple pixels at once.

Personally, I would consider continuing to scan to PDF, but with the highest resolution possible. Most copiers today will scan to PDF, and you should be able to use a copier at an office supply store (for a price) to scan images to PDF. Alternately, once you have the image in a picture file, computers will either natively, or utilizing a program scan your pictures to PDF.

Of course, as already mentioned, IMSLP is an excellent public domain source for music published before 1925.

Michael

[Warrentdo]

Hello Michael,
the purpose of my scanning is to see if I can get a crisp scan. I need something that looks like paper and not something that i struggle to read.
In my test Jpg are showing to be rather large. 1200dpi with encoding as standard 7846kb and that had gray scaling even thought the scanner was set to black and white. 1200dpi with encoding to progressive was 6511 kb but still had slight gray scaling. when I make a PDF (with Epson scan) i also get more grayscale. the only way I have found to make a 2 bit PDF is using Microsoft PDF print driver. and that sends the size through the roof which I find strange. Do you have an example of a scan that you have done?
Warren.

[myorgan]

Warrentdo,

When I get to my other computer I can send it later tonight. I have noticed PC and Mac handle file sizes quite differently. I've had MUCH better luck printing to a PDF on a Mac vs. PC. The file sizes are worlds apart with no difference in quality. On the other hand, I've found when re-sizing or converting a filetype from TIFF to JPG, the PC has much better compression than the Mac. Consequently, I go between the two-depending on the task at hand.

I've also noticed when printing a PDF, the PC is much quicker than the Mac when printing to my HP LaserJet. OTOH, the Mac is much faster when printing via IP to our Toshiba copier at work.

When doing OCR (or MCR in your case), I found the PC had the best quality scan from a TIFF file than the Mac. So, it depends on your end product what you use to obtain that result. You're doing the right thing by experimenting.

Michael

[smithge31419]

Just a couple of comments about what I've used for successfully scanning sheet music. Back when I started doing this, there weren't too many options available. For scanning we were pretty well stuck with the scan software that came with our scanners. For music OCR, Photoscore was about the only one that really worked, and for editing, it was either Finale or Sibelius. I settled on Photoscore and Sibelius, since Photoscore seamlessly integrates with Sibelius. Were I to start over, I might choose something else, but that is what I spent my $$ on, so that is what I've got.
A couple of things that I have added to my setup that has made life much easier is a full size (tabloid or A3) scanner and full featured scanning software. Sheet music rarely fits on standard letter sized paper, and good software lets you scan to multi-page pdf files easily. There are several good tabloid sized all in one printer/scanners available for under $250. I happen to be using a HP Office-Jet 7740. For scanning software, I found Vue-Scan back when I could no longer get Windows 10 drivers for my old photo scanner. It is a bit expensive, but it runs on Windows, Linux, and Mac and supports just about any scanner that has ever been made. (and you only need one license)
Some comments on use: Vue-Scan can be a bit complicated to get set up as it allows you to configure just about everything you might need and quite a bit of other things as well. Once you have done this though, scanning is quite simple. Place document on Scanner, click Preview, adjust crop lines, click save. For subsequent pages do the same, except click save+. I save all my music files as 300dpi 1bit B/W PDF's. Quality is plenty good for reading by Photoscore or for printing. I have attached an enlarged JPG image of a snippet of a 300dpi scan. This snapshot is from a 6 page piece of music. Total file size for the 6 page PDF is 365 Kb. Note: If I had saved it as "grayscale" the file would be much larger.
Regarding Photoscore, I find that it works fairly well. There are some notations that it will NOT recognize, and there are almost always a few errors that must be cleaned up before you save the file. Old, worn originals with lots of pencil marks and erasures are particularly troublesome, although the newer versions of the program are much better than they used to be.

[myorgan]

I wish I could like this post more than once!!! Stated very well.

Michael

[Warrentdo]

This is crazy, after more messing and examining with various tools the PDF's that I have been creating are 1bit!
It was the viewer that made it look like 8 bit with dithering!
Just something to watch out for -
Viewing through Microsoft edge added extra smoothing bits to the screen that I dont want Capture ME.PNG.
Viewing through Adobe reader viewed just as it should Capture AR.PNG.
Warren.

[myorgan]

Warren,

The other option you have is to sharpen the images. However, sometimes it will cause OCR to read things differently if something you have sharpened is closer to white than black.

I'm glad you got your issue sorted out.

Michael