Importing printed sheet music into MuseScore

• Oct 1, 2020 - 15:52

It is possible to convert printed sheet into MuseScore (.mscz) file format for playback and editiing.

You will need a music-scanning program like SmartScore (https://www.musitek.com).
SmartScore converts scanned and/or PDF image files to the MusicXML format which can be imported into MuseScore. Imported scores appear as the original printed sheet music and are fully editable in MuseScore.


Comments

This is a prpriatary commercial software though and costs 79$ to 399$ depending on version and as such your post comes across quite spammy. And it is not available for Linux.
Audiveris does this too and is free and open source

In reply to by Jojo-Schmitz

There are quite a few "music-scanning" applications out there both commercially available as well as free and open source. MuseScore users who are looking to convert and import sheet music should be made aware that the technology is out there. And by all means, they should be able to test any or all of them free of charge and decide on their own if one or the other program serves their purposes.

The Linux version is under development according the folks at Musitek.

In reply to by Jojo-Schmitz

I have uploaded 4 PDF files (simple to more complex) to the MuseScore PDF conversion page (www.musescore.com/import). Processing was quite slow and eventually, each file failed to convert (Unsuccessful)... this an even stronger argument to interested folks that they should carefully test scanning programs against one another before committing. Finale used to include a light version of SmartScore and was about to add a newer iteration with the release of Finale24. They posted a video showing an accurate conversion from a mobile camera to Finale (via the SmartScore plug-in) which caused an uproar among publishers and composers who came down hard against it. Finale was forced to pull it out before Finale24 was released... not because it wasn't any good, but because it was too good.
https://www.scoringnotes.com/opinion/makemusic-pulls-pdf-importing-and-…

In reply to by Kittifer

And Jojo's point was that you are advertising a product which is not permitted in the forums. I'll let the administrator decide if your posts are considered spam unless you continue to advertise your product, in which case I will report it and you as spam.

In reply to by Jojo-Schmitz

My point here is that MuseScore users ought to be aware that the technology is out there and that testing one system against the other (whether free or commercial) is the best and only way to decide whether or not to spend time and/or money on the effort.

In reply to by Jojo-Schmitz

I would like to confirm here that the use of the MuseScore import has, for me at least, been totally useless. I have tried 7 or 8 uploads, and all have either been "unsuccessful", or have been illegible garbage. I will try the direct use of Audiveris, with minimal hopes. But the sheer volume of similar software out there is completely overwhelming and utterly confusing. Can you (or anyone) suggest a simple, relatively inexpensive
pdf-to-musicxml converter other than Audiveris? I don't need bells & whistles; all I need is to get my pdf's into MuseScore in a usable form. Thanks for any and all suggestions.

In reply to by tnjazzgal

"Can you (or anyone) suggest a simple, relatively inexpensive
pdf-to-musicxml converter...?"

a) Download the Windows version of Audiveris, and test it by changing various parameters. That will cost you nothing but your time.
b) Accept the fact that commercial software for optical music recognition (OMR) is expensive. There are no simple and inexpensive options which work really well.
c) All OMR systems have quite a steep learning curve before they work well (95% correct or better). You have to read and re-read the manual! Patience is required.
d) The only commercial OMR solutions I know from regular personal use are these:
- Sharpeye: old-fashioned DOS-like interface, very good at music recognition, not so good at lyrics, reads TIFF (and BMP) files but not PDFs, never crashes (YMMV)
http://www.visiv.co.uk/
- PhotoScore: complicated but effective interface, good recognition of both music and lyrics, reads PDF files, occasional crashes/errors
https://www.neuratron.com/photoscore.htm

In reply to by jeetee

"One that got some decent remarks in the facebook user group of MuseScore is the PlayScore2 app: https://www.playscore.co"

Thanks for this tip.

I downloaded this and tried it out. Initial results are impressive. The score plays. But with a few caveats.

  1. The import from PDF left all my PDFs grayed out in the Finder window. I'm not sure why or if that's due to the trial version. But the developer took the PDF I emailed him (the same one I posted below) and imported it for me and sent me a Playscore file in return.

  2. The file displays exactly like the original. But the display appears to just be an image of the original. So it appears that it will display even content that it doesn't recognize. It's not like MuseScore which renders all its content from scratch. But the playback cursor follows that image and plays what it sees.

  3. Playback played the whole score. Impressive

  4. Export to MusicXML: this appears to be disabled until I pay. I have asked that he run a sample file for me to see how it will look in MuseScore after the data has been serialized and then unpacked by MS. But given how it sounds, I'm fairly optimistic.

  5. For Apple, this is an iPad app. But the developer has enabled the ability of the app to run on Apple Silicon Macs, such as my MacBook Air M2. I'm using it right now. This is a huge productivity boost, since it will be able to run alongside MuseScore, rather than having to switch between devices.

So far this is the only product I have found that can read this file. I'll try to keep the thread posted with further results.

Playscore import.png

In reply to by reggoboy

UPDATE:

The MusicXML export is not available in the trial version of PlayScore 2. So the developer ran this file for me, which I have now imported to MS. Below is a screen shot of the same passage that I showed above from the PlayScore 2 GUI.

As you can see, it had no problem with the gray lines, and it did a pretty darn good job with accuracy, especially considering that all the alternatives I tried failed completely or very nearly completely.

Software to the rescue, doing what it does best :-) Now I have to decide if I'm willing to pay for this.

MuseScore Import.png

In reply to by Jojo-Schmitz

THANKS! but this audiveris just isnt working. i mean the programs all right, but saving my exported files is literally impossible. it only saves it after about 20 clicks and never in the directory ive chosen it to be in. also i cant even use those exported saved files in musescore, there is always some error f.e. unexpected measurein line 32 and something like that.

I'm going to assume this is all in good faith, but frankly, this is the forum for MuseScore, not for other, only tangentially related software.

If MuseScore users want to be aware of the options for music-scanning software, they need only perform a quick internet search.

I appreciate that you took the time to post here, but I'd advise deleting this post, since it's really just off-topic.

edit: I see that this is your product, making this post simply an advertisement. In that case, this definitely goes against the spirit of this forum, and I would strongly suggest you delete this (it'll probably get removed anyway).

In reply to by TheOtherJThistle

You are correct. My main purpose is to clear the air regarding music scanning and PDF import into MuseScore. I have searched musescore/community and other forums and found so many layers of confusion and disinformation regarding this technology that I am forced to speak out. See:
https://musescore.com/groups/improving-musescore-com/discuss/5031824

Music recognition technology has developed to such a level today that there is no argument whether it works or not. You are also correct that I am associated with the company that publishes SmartScore. I encourage all MuseScore users interested in converting their sheet music (whether scanned or in PDF format) to investigate and rigorously test what is out there today. Here's a place to start (Wikipedia):
https://en.wikipedia.org/wiki/Optical_music_recognition

In reply to by Kittifer

You might not be aware, but there is a distinction between the .org and the .com site (see https://musescore.org/en/faq#faq-20657) and it might be better to continue/keep discussions from one platform on that same platform.

I have no doubt that you continue to develop your product, just like your competitors do as well. I also don't contest that you've likely made major improvements, just like they did. But there is a difference between answering to a relevant topic and creating a topic just for the purpose of advertising your product. Without mentioning the competition, such a post becomes harder to regard in the spirit of "encouraging all MuseScore users to investigate and rigorously test what is out there today".

There is no issue with mentioning your product or the advanced made in it at all, but do it in a relevant place.

Now here's the fun part:

If you read linked topic (apparently the reason for you to be forced to speak out) then you might've come across this particular comment. It explicitly mentions your product.
Or the comment right below it, also evaluating your product.

Now I'll admit that those aren't the most favorable comments about your product. And given that they're from over a year ago, likely (and hopefully) your product has improved since then.

Which brings me to my two main points:
1. Obviously the users are aware of your product, so this post is not just about informing them that your product exists.

  1. If that topic (even if it's just an example for one of the many discussion that happened in the past) compelled you into replying; Why didn't you?
    Why not simply hit that reply button right there? Invite those that have tested your product x time ago to redo their testing; show them what/how you've improved? Perhaps inquiry them for sample scores that didn't work before and show them the current result of using your current version?

In reply to by Jojo-Schmitz

Since you have removed our shoot-out challenge specifically requested by jeetee, I suggest you, jeetee or someone else post a list of Music-OCR/OMR, Music-scanning, PDF-to-XML programs out there and invite interested MuseScore users to download, test and compare respective demo versions. I'll be stepping away from this discussion. Cheers.

In reply to by Jojo-Schmitz

"The folks at Musitek" are SmartScore developers. They are contractors, not employees. They call the shots as to when and how a project is completed. I'm curious... what is your motivation Jojo? Root out potential competitors of MuseScore? I would have thought promoting a TECHNOLOGY (which remains misunderstood by a vast number of folks) that works hand-in-hand with MuseScore would be a boon to your user base. Feel free to delete this thread and any mention of alternative technologies other than Audiveris.

In reply to by Kittifer

I can't speak for anyone but myself, but here's my take:

First, SmartScore in no way competes with MuseScore. It's a technology that can add value for MuseScore users, just as Audiveris can, with different plusses and minusses. We here on this community-driven forum are happy to see information posted that can help MuseScore users discover and use technology that is potentially useful to them. And it's fine by me for employees of the companies providing these technologies to be providing some of this information. But, since it is a community-driven forum, I think it courteous and best for all concerned if people are upfront about their connections when doing so.

In reply to by Jojo-Schmitz

Mea culpa. I am associated with Musitek and have said so in this thread. It's true that I didn't state that fact at the outset since there is much sensitivity that a company might use this platform to "advertise" their products and therefore, risk that the topic, as important as it is, would immediately be removed. Yes, I included a pitch for SmartScore. My intent remained the same... to promote music-scanning technology for those who might not be aware that it exists or who misunderstand it. I probably should have announced that I am associated with Musitek and will proceed to pitch SmartScore 64. But I didn't. So call it "spam" and delete the thread. Or sit tight and let's see the results of a music-scanning shoot-out. We'll take the risk of being eviscerated by folks who find SmartScore inferior to its competitors.

In reply to by Kittifer

Again, I can't speak for anyone else - and I should clarify that while I am one of the many volunteer developers of MuseScore, like almost everyone else who can say that, I don't "work for" the company. But anyhow, as far as I am concerned, there is still good information here in the thread, so I wouldn't be in favor of seeing it removed. If it were just generic advertising copy posted week after week that would be another matter - that would seem like spam. But I'm happy to see you continue to contribute information of value here. It just feels better to have things out in the open a little more.

I wish you success, and I hope your product serves the needs of interested MuseScore users well!

May I add one thought to this music-scanning story, for anyone who might be interested?
I own a music-scanning software (it is not important which one) and it works quite good.
However, a musical score can be and often is very complex (even a piano piece, let's not mention the orchestral score). Thus, you have to edit manually many, many things even before you end with the correct MusicXML which you can import in MuseScore and continue with the editing. I do not speak here about some smaller layout issues, but things like the correct total rhythmic value of measures. This editing have to be done using the user interface that is provided by the music-scanning software. The quality/usability of this user interface, of course, is much lesser that the quality of the MuseScore user interface. And that is the weak point.

There is no need for additional software if you have a scanner, which most desktop printers today have built-in, and 99% of those scan to .pdf as well as .jpg and other image formats.

Upload the .pdf to https://musescore.com/import?from=upload which uses Audiveris software. The .pdf file is converted to the Musescore .mscz format for download to your hard drive or wherever you want to park it. Although I've not converted many files, yet, I have had no problems with the files I scanned on my printer and converted via Musescore - 100% success with no glitches whatsoever.

No need to "import" anything into Musescore this way because .mscz files are native to Musescore, so the user simply opens it from where ever it was saved.

I, too, see the original post as spam, so there is no way in hades I'd click the link or even look it up in a 'net search much less buy the product. I can guarantee, though, that if ever involved in a discussion in which the product/site is mentioned, I will add my low opinion to the conversation. Shame on you.

In reply to by Ohmo

"There is no need for additional software if you have a scanner"
Have you actually used the MuseScore option File > Import PDF... for any serious amount of optical music recognition (OMR)? Sorry, but my experience of trying this link has been miserably unsuccessful. In an effort to be fair, I even downloaded a copy of Audiveris in order to test adjusting its various parameters on my PC.

On the other hand I have used two commercial OMR programs with considerable success, scanning and converting to MusicXML more than two thousand pages of Public Domain scores over the past ten years.

So frustrating. I have tried every piece of software I have found recommended online, and can't get anything to even come close to reading the attached file (or ones like it).

I have tried masking out extraneous text / symbols, then saving as PNG, then converting back to PDF...

I have tried using macOS Preview's Quartz filter to convert the gray staff lines to black (doesn't work)...

I can't get anywhere with it, even though it's crystal clear to the eye.

Any suggestions are greatly appreciated!

Ready For Love - Piano RH.pdf

In reply to by reggoboy

I can only repeat my standard advice: time spent entering the music into MuseScore will almost always be more productive than time spent trying to find software to do it for you and then correcting the resulting errors. Well, I've never necessarily worded it that way, but it's as good a way as any. There is no way that entering that music should take as long as an hour, so if you've spent more than that, it's time to cut your losses and just enter the normally.

In reply to by Marc Sabatella

Thanks, but your response is very short sighted. I just provided this as a sample. This song alone has 2 more files (not shown) for other parts that have to be merged together. And then there is the ceaseless flow of other songs that I'm handling. Doing all of them by hand is impossible. And as I've said before, if we're going to do everything by hand, we don't need MuseScore either; we can get out our inkwell like Bach and do it all by hand. The point of the software is to help us. If it's not going to help, it shouldn't exist at all. But I'm not convinced that in 2023 there is no way to read a crystal clear file like the sample I attached and give me MusicXML.

In reply to by reggoboy

Sorry, the staff lines are too thin and too light. No PDF reader can read this. PhotoScore can't begin to read it. Changing the contrast or anything won't help. It doesn't look like notation software produced it. I think real musicians would have trouble with this. Compare this to almost any printed music.
Maybe if you took a ruler and fine point sharpie to it, it "might" work.
Don't believe me? As an experiment, enter a page of this in MuseScore. Then save it as a PDF. Try your PDF readers and see what you get.

In reply to by bobjp

"Sorry, the staff lines are too thin and too light."

Yea, I've been suspecting the gray staff lines. On macOS, everyone says that if you export to PDF using the Quartz filter "Black and White", it will fix that, but it just makes the staff lines go away completely.

"Changing the contrast or anything won't help."

Why not? Something that would darken gray to black might fix this? The color palette here is very flat: black, white, and gray.

" I think real musicians would have trouble with this. "

I like to think of myself as at least a "partially real musician", and I'm fine with it visually. The notating of the timing is annoying. But if it's legit, a software program should handle it. Unless it doesn't like things like that [...9...] run??

In reply to by reggoboy

That's just it. Your PDF is not a standard notation look. I can read it, too. But not happily. I have a paid PDF reader. It refuses to read this. Also, this looks like a DAW's attempt at notation. Many DAW's try to interpret octaves and rhythms the way we see here. On a single staff. What is the source of this?
This is not a reader or notation software problem.

In reply to by bobjp

" Also, this looks like a DAW's attempt at notation. "

Yes, it's definitely done by machine, as another reader also noted. But that's the direction of the industry these days. Reading hymnals and paper scores is an old problem now, and it's been solved. More and more content for musicians online is generated by machine. Ideally, those providers would provide more than just PDF, but rather the raw XML. But often they do not. So this is a problem needing solving.

Fortunately, computers are particularly good at reading other computer data, and I'm having early luck with PlayScore 2...

In reply to by reggoboy

I just need page 2 of the score you've been showing. Depending on the format you can zip it and post it here. My PDF reader crashes every time I try to read the one you posted. I've read plenty of PDF's with it and it never crashes. It also doesn't like the setting it was scanned at. Too low. It shows up with a light tan back ground instead of white. If that is accurate, it could be the problem. Who knows. Thanks. This is how we learn things.

In reply to by bobjp

Okay, it took some hunting, but here it is. After this failed Audiveris, I proceeded to "mask out" all kinds of extraneous text and marks in the PDF, and then exported as PNG and back to PDF, hoping that would hide things. But I'm not good at reading PDF code, so I don't know how effective those efforts were. And they didn't help the processing with Audiveris. I do think that it was my scrubbed version that I fed to PlayScore, though. It would be interesting to try the raw version with PlayScore to see if it would have been as successful. But I'd probably have to pay for the product at this point :-)
READY FOR LOVE - PAGE 2.pdf

In reply to by reggoboy

I tried your sample PDF page (READY FOR LOVE - PAGE 2.pdf). I used two commercial software packages for optical music recognition (OMR). In both cases the import failed completely, because the barlines and stave lines are too thin and are grey (not black). These basic elements were invisible to the software, and as a result no music was captured at all.

Sorry, but I think you have zero chance of OMR success with such poor source material as the starting point.

In reply to by reggoboy

There is just something basically wrong with this PDF. I wish I knew more about image files. I have PhotoScore Lite that came with Sibelius. There are two steps to reading a PDF using it. First the file is input, then it is read. Every other PDF I have tried automatically goes directly to the read part and is ready to edit. This PDF gets stuck on input. It sort of tries, but is sideways. When I try to correct it, the program crashes. If I change the page to a tiff, The program wants to know the DPI I want to import at I choose 300. The preview shows plenty dark staff lines, but still sideways. The software still won't read it.

In reply to by underquark

"Something has already had a go at "importing" this, maybe from an .MP3 file and it has generated some odd results. Even if you could enhance the near-invisible stave lines, you'd still need to manually correct the notation to make it more conventional."

Yea, I'm not thrilled with their notation. That's their lousy way of notating grace notes.

In reply to by underquark

"There are no grace notes in the image I posted"

Yes, but those 32nd notes visible in your screen shot are really grace notes in the recording. And my point is that rather than notating them as such, this publisher's software renders them like this, which is pretty lousy, and I'd have to fix those. Still, with the amount of material I get from this publisher and need to move to MuseScore, it would help to have an import tool. If you're interested, see my comment above about optimistic early results from PlayScore 2...

In reply to by underquark

"OK, so your publisher seems to be the main problem, not MuseScore."

Unless we get technical logs from Audiveris clarifying WHY the import failed, we are not in a position to make that evaluation.

The publisher could have made the gray lines black. And they could have avoided 32nd notes and "(9)-tuples". All of that could have made Audiveris' life easier. Or maybe that had nothing to do with the failure. Either way, I don't see anything wrong or ambiguous about it the original score. This is not some fax or a handwritten score.

We need to stop making excuses for software. When software performs badly, the solution is not to make the problem easier; it's to make the software better. Google never told web sites to clean up their content. It figured out how to interpret what it was given. That's how good software developers think. I've been writing software for >40 years. And my Dad (rest his soul) always said: "I don't make excuses; the buck stops here."

Anyway, all this blaming the original publisher is moot now since I have shown above another product that imported it fine.

In reply to by reggoboy

It’s certainly not impossible to enter music by hand even if there are two more pages. But it is true that at some point it might turn counterproductive. You haven’t provide enough info for us to understand the trade offs. If we talking tens of thousands of pages, then it might take thousands of hours, and so spending hundreds of hours finding the perfect PDF import program and the import and then hopefully only hundreds rather than thousands more fixing the errors will pay off. That’s just not been my experience though.

You analogy to doing things by hand makes no sense. Writing out music is extremely difficult and time consuming if you want professional results, and you won’t get playback. Entering music into MuseScore is far faster, gives far better results, and provides playback, transposition, part management, easy editing, and tons of other benefits that writing by hand lacks.

In reply to by Marc Sabatella

"It’s certainly not impossible to enter music by hand even if there are two more pages. "

It's impossible because I have a finite amount of time. I have 24 hours in a day, and so it's physically impossible for me to keep up with my workload by doing this by hand. But I don't know why I have to explain that.

"You haven’t provide enough info for us to understand the trade offs. If we talking tens of thousands of pages, then it might take thousands of hours... "

It doesn't matter. You don't need to know that detail. I know how much time it will take me to do it by hand, and I will do that when I determine that the software is working against me. The same is true for every other person commenting here; they're adults. They can figure out when to cut bait; we don't need you to tell us.

This thread is to find software that helps make this process helpful. It's NOT helpful to say "do it by hand". That comment has no place here. It doesn't help. It's off topic.

In reply to by reggoboy

Knowing how many scores you have to do is relevant. Think about it. Let's say you have one page of music to enter that could be done in an hour, and it takes you 10 hours of searching to find software that will allow you to scan it successfully enough that it only takes half an hour to fix the errors. That's not saving time - one hour versus 10.5 hours. If you have 10 pages to enter, you still come out behind - 10 hours of search plus 5 hours of scanning and fixing, versus 10 hours to enter normally. If you have 40 pages of music to enter, then you finally break even: 20 hours of searching for software, 20 hours to enter those 40 pages is 40 hours - same as just entering the music normally. Only if you have more than 40 pages to enter, then it starts to look like a win, if indeed you are expert enough using MuseScore that you can scan and fix up a page in only half an hour.

The point is, we absolutely want to save you time, but without knowing these numbers, it's impossible to say where the tradeoffs are. And the reason our insight is relevant is, we have a pretty good idea of the kind of results are typically obtained by scanning software and the amount of time people typically spend fixing errors. I don't know enough about you to know how expert you are, or how familiar you are with these issues, so I'm just giving the advice that probably applies to 90% of users. Maybe you are indeed in the top 10% percentile, and if so, I salute you!

Anyhow, sorry if you aren't wanting our help, so by all means, feel free to ignore our wisdom. Hopefully you do indeed manage to figure out ways of saving time on your own. Good luck!

In reply to by reggoboy

"The same is true for every other person commenting here"

Not for me. I am very grateful for this free software that helps me write/transcribe pieces quickly and the result looks professional. The software does not work against me, on the contrary - otherwise I would not like to use it so much.

In reply to by Pentatonus

"Not for me. I am very grateful for this free software"

I'm also very grateful, and I mention that frequently, and contribute feedback as much as I can and am a paying member of the .com community and the (now free) mobile app, When I say "The same is true for every other person commenting here" I'm referring to the fact that everyone is capable of deciding for themselves when it's more work to fight the import failures and just enter the content by hand.

In reply to by Jm6stringer

"The tuning is explicitly given as (standard) tuning: E A D G B E, but many notes are lower and consequently unplayable:"

Good eye :-) That header is just boilerplate. The software that produced this piano score also has other "layers" showing the guitar parts, not included here. I guess it's a bug that they didn't hide that "EADGBE" header when displaying the piano score. But obviously, all these notes are fair game on piano.

I have been working on importing PDF music from choir that has SATB and accompaniment so I can separate the parts for practice. I used Audiveris on Windows and Linux and I am a fan. It is quick, and once I figured out how to make sense of the pink areas and entering symbols I have found it efficient. Once it is imported to MuseScore I can clean it up the rest of the way and edit the voices. I love how all of this is FOSS as well.

Do you still have an unanswered question? Please log in first to post your question.