Why doesn't MIDI output work on Musescore 4?

• Jan 6, 2023 - 17:07

It used to work perfectly with the previous version, but now it doesn't work. I double-checked to see if it was a problem with my digital piano or computer, but it isn't, because MIDI output works everywhere else.

Comments

• Mar 11, 2023 - 08:57

I have the same problem.
The midi input/output worked initially, but no longer with the current version v4.0.1.230121751.
Here is a snapshot of my preferences, where I have selected my keyboard as input and output.
When I play the score, it still goes to my PC Loudspeaker instead of my MIDI keyboard.
I have also "Activer input MIDI" in the playback setting, although is should not affect playback, I guess.
Is there any other setting that I should change?
M4_midi.jpg

Note: initially, the selection was totally empty; the pulldown menu was populated with "Aucun périphérique" and "CASIO USB-MIDI", but the latter could not be selected. I found a workaround by first selecting "Aucun périphérique", and then "CASIO USB-MIDI". This is probably an unrelated small bug.

UPDATE: when I set the volume of my CASIO keyboard to maximum, I can hear the notes very weakly. There is also a note (A) superposed to each of the actual notes.
Midi input did not work half an hour ago, but now it does; could this be linked to a test I did with Musescore 3 to confirm that the keyboard worked fine with M3?

Reply

• Dec 23, 2023 - 20:33

Same bug in 4.2: Note on message is bad issued so the velocity is set to 1 or 0. Any hack like using MIDIOX as a midi transformer, could make it sound, but with no expression since volume will be loud or soft depending on the values given to 1 and 0 on traslation. This is unnaceptable and a pretty dumb bug for a released music software.

Reply

• Aug 20, 2023 - 03:37

I can confirm the same behavior: MIDI output does not work properly. If I turn up the volume on the Casio keyboard (with built-in USB port, shows up as "CASIO USB-MIDI" in Preferences), I can barely hear some notes playing, even when there are no notes (only pauses) in the staff. And by barely I mean about the same loudness as the speaker background noise.

MuseScore 4.1.1 on Manjaro Linux with KDE and Pipewire. I don't see any MIDI connection in qpwgraph, which makes me believe it's trying to connect to MIDI directly instead of using JACK/Pipewire.

Some related issues on GitHub:
* https://github.com/musescore/MuseScore/issues/16787
* https://github.com/musescore/MuseScore/issues/18382

Reply

• Nov 27, 2023 - 09:00

I tested again today, with the latest version 4.1.1.232071203 Révision e4d1ddf, on a desktop PC running the latest version of Windows 10.

A good point for version 4.1.1 is that PC playback issues (cracks and other perturbations) are seemingly solved.

For Midi, I have still the same problem as before: Midi input works, but not Midi output, despite performing all selections mentioned by Marc S. Sometimes, I only hear a repeated tiny A4 note in the background, as already reported by other users.

Sequence for testing:
1)- launch Musescore 4
- create a new score, just selecting keyboard/piano/add
- edit preferences i/o: midi menus are empty so I click on the pulldown and try to select Casio midi, to no avail.
using a trick already mentioned last year, I first select "aucun périphérique" (no peripheral) and then reselect "CASIO USB-MIDI" (I already signaled this bug on the pulldown selector on the previous version: it is still there).
so I finally get CASIO USB-MIDI for both Midi input and Midi output.
- check that the playback menu has "Activate Midi input" selected
- click on the first rest and press N
- play notes on the Casio keyboard => the notes appear on the M4 score. => Midi input is workging
- press N to exit the input mode
- click on the first note and press space to play the score => it only plays on the PC Window, not on the Casio.
2) close and restart Musescore, opening the last played score:
- the CASIO USB-MIDI selection and the "Activate Midi input" are still there
- but the Midi output still does not work.
3) Launch Musescore 3.6.0, opening the last recent score
- press space => the score plays on both PC desktop loudspeakers and on the Casio Keyboard.

Conclusion: I will once more stick to 3.6.0 until a new release of M4 is available, hoping that these issues will eventually be solved.
It's a pity because I appreciate the new layout and menus organisation of M4 and I am sure that a lot of good work has been done by the developpers.
Unfortunately, mscz 4 files incompatibility makes it cumbersome to move back to M3 once you edit a file in M4.
Therefore, I do not dare to create (or even edit) my scores in M4 until basic functions like midi playback are reliably working.

Reply

• Jan 11, 2024 - 03:33

I figured out a different way to connect MS4 to midi connections for each instrument in a score to support the following use cases:
1- Connect to non vst3 plugins
2 - Connect instruments to vst/racks in your DAW
3 - Connect instruments to external midi hardware

Using the Element plugin from kvraudio installs a vst3 plugin that can be added to each instrument in the mixer. The Element plugin allows the creation of a workflow where you have multiple options to route and monitor the midi signal sent by MS4. Drag the midi port for your instrument to the editor and link it to the midi input - that is it.
I was cable to configure a piano and a drum track going to a Yamaha P125 and a drum VST in Reaper, properly equalized. When I press play in MS4 the piano plays on the Yamaha and the drum plays on a Reaper track. The DAW is irrelevant. As a side note I tested Dexed (DX7 emulator free) in MS4 as piano VST and was very pleased with the results.

Attachment Size
elemen1.PNG 6.72 KB

Reply

• Feb 7, 2024 - 16:49

Tested again today with version 4.2.1
Something new is that I can hear the notes of my score on my midi device.
Previously, it was simply a repeated note.
Bad news ist that the volume is again very very low, barely audible with volume set to maximum level on the device. Changing the setting of the mixer master has no effect on this.
I also noticed that only one note of the chords is played on the device while the full chord is played on the PC loudspeaker.
Just sending this update in case it could help to localise the cause of this issue.

Reply

• Feb 11, 2024 - 06:53

I have the same problem with my Yamaha Arius YDP 143. I can input notes to a score with it, but musescore doesn't output to my piano even though in I/O my Arius is recognised for input and output. So also here I'm forced to keep using version 3 because there it's working fine.

Reply

• Mar 29, 2024 - 12:54

This problem is caused by all note on/off velocity value generated MU4 is 1. It's too small value.
MIDI velocity can take value from 0 to 127, usually around 64.
This occurs in all Ver.4 series. This does not occur in Ver.3.
It's very inconvenient.

See this figure.
https://musescore.org/sites/musescore.org/files/styles/width_740/public…

https://musescore.org/ja/node/338263#comment-1159523

I and some users reported same issue in Japanese forum.
We waiting fix this for a long time.

「[MS 4.1.0] Note On/Off Velocity がすべて 1」
https://musescore.org/ja/node/352572

「[MS 4.0] Note On Velocity がすべて 1」
https://musescore.org/ja/node/338263

「MIDI OUTの不具合」
https://musescore.org/ja/node/340226

Reply

• Apr 9, 2024 - 04:51

Did you check the velocity of the MIDI messages as indicated above?

In any case, to be clear: museSvore is not “MIDI music software” - it is music notation software Thant just happens to employ MIDI for its own internal playback. MuseSvore having limits in its MIDI I/O is no more surprising than your favorite DAW having limits in its music notation.

Reply

• Apr 11, 2024 - 23:45

Sure, but it worked fine in version 3, so it is a feature that have been spoiled somehow during development, so it shoudn't be so hard to fix it, since it is a problem that during building the midi message, the 2nd data byte in NOTE-ON message, which is the note velocity is set either to 1 or 0, instead of the real value. Moreover, it works fine when using the internal synth, so either internally you ar not using standar midi messages or the routines that creates the messages are different for internal and external synths. Attached you will find the Note-on midi message format I mentioned.

Attachment Size
Captura de pantalla 2024-04-12 005253.png 26.02 KB

Reply

• Apr 12, 2024 - 11:54

Fully supported.
It is hard to understand why it takes so much time to simply recognise that there is a bug.
I would like to ask this question: are there users (running on Windows 10) that do NOT experience this bug ?

Maybe this forum is not the right place for signalling bugs in v4.
I have now posted a bug report on github.

Reply

• Apr 12, 2024 - 23:31

https://github.com/musescore/MuseScore/issues/22354#issuecomment-205254…

By captured MIDI output, I may have found other problem.
This sample score is contain 4 notes, C4, E4, G4, C5.
The corresponding MIDI messages are as follows:

C4 note on/off: 90 3c 01/80 3c 01
E4 note on/off: 90 40 01/80 40 01
G4 note on/off: 90 43 01/80 43 01
C5 note on/off: 90 48 01/80 48 01

But the captured MIDI output also contained the following note on/off messages:

E5 note on/off: 90 4c 01/80 4c 01
F5 note on/off: 90 4d 01/80 4d 01

What are these? How about others?

Attachment Size
MU4_MIDI_out.png 27.2 KB
MIDI_out_test_CDEG.png 16.27 KB

Reply

• Apr 13, 2024 - 13:37

Thanks. I also confirmed it.
I guess from the results of your experiment, these MIDI messages may be for metronome sounds.
In the current implementation, it cannot be turned off... ．
I think this is a bug too.

I think that MIDI messages for metronome sounds should not be output.

Reply

• Apr 13, 2024 - 14:13

Yes, I agree with you, taking into account that they are sent in the same channel (1) of the instrument, in case they sound, piano notes will sound instead of metronome sounds. And what about those pitch bend and sustain pedal events? Repeating this events with 0 value is useless and can eventually cause that a sustain pedal maintained between measures be cut too early.

Reply

• Apr 20, 2024 - 20:00

To have music notation software that "is not MIDI music software", in 2024, is like having a car that is not meant for driving on a road. I'm sorry. That sounds harsh. I mean, if you're going to support control of musical instruments, which is a great idea from notation software, you need to do it adequately. If it's going to be poor, don't do it at all.

Reply

• May 30, 2024 - 17:16

I can think of three (no, apparently just two) examples of "cars that are not meant for driving on a road": a) drag racer, b) full-time 4-wheel drive, and c) well, I had a third example in mind a minute ago. Doesn't make them "inadequate" or "poor". Yes, both can be driven on a road, but they are not meant for driving on a road.

MuseScore is intended to create standard European music notation. If it also does more than that (and it does, a lot more), consider yourself fortunate.

Reply

• May 30, 2024 - 17:46

One of the big issues is that MS3 had this functionality so there was a reasonable expectation that MS4 would be at least as good in this area, if not better.

Notation software has supported MIDI since the 1990s so without it in 2024 is like a car with 3 wheels: there just seems to be something basic missing! (I know that there are 3 wheeled cars but just try driving one around a tight bend.)

Reply

• Jun 2, 2024 - 11:37

I thought that would be implied. Hard for MIDI to be usable without notation software existing.
Anyway, here is graphical notation software, supporting MIDI, from 1988 on Atari ST.
https://www.muzines.co.uk/articles/c-lab-notator/3970

I had a PC at the time. I used Cakewalk for DOS, which was text-mode only and had piano roll,. It supported MIDI.

I recently restored some DDS-4 tapes from the 1990s and the contents includes that 1980s version of Cakewalk. The readme is all about MIDI. Actually, better than that, I just ran Cakewalk Pro under DOSBox staging, and played a MIDI file with Munt - an (MT-32 emulator, running under Windows. Worked perfectly fine. The MIDI file I just played was called BACH1-32.MID , BWV 784, using a harpsichord patch. I recorded this piece on acoustic piano just this year. There was at least one pitch mistake in the MIDI file. Not sure if it came with Cakewalk or not. I need to record it on acoustic my harpsichord also. Just have been too lazy to tune it.
I tried a few other MIDI files from games - one was too large, the other didn't play Sysex correctly. Not too surprising an old DOS program could not handle these.

Reply

• Jun 2, 2024 - 13:27

Re: "I thought that would be implied" : that seems a strange inference to make from what I posted. Marc clearly understood what I meant.

It was also easy for MIDI to be usable without notation software. What about MIDI sequencer software? What about MIDI control keyboards?

That's the same link that I posted so I think you have not read my response properly either.

Reply

• Jun 2, 2024 - 14:09

FWIW, while this is a bit of a tangent, I rather doubt notation software was even a twinkle in anyone’s eye when MIDI first came to be in the early 1980’s. MIDI was developed for digital keyboards and synthesizers to communicate in real time for live performance / recording. It was and is used extensively by many with no reference to notation software or indeed computer software of any kind - just direct cables connecting standalone keyboards and synths. Of course, it wasn’t long because people started realizing it could also be used for other things. But it’s not really accurate to say “hard for MIDI to be usable” without notation software.

Anyhow, while at this point 40 years later, most users of notation software are playing back exclusively via software synthesizers and virtual instruments like Muse Sounds and VSTi rather than via external MIDI devices, no doubt some percentage of MuseScore users do still rely on MIDI output. So, hopefully the issues will be fixed. As always, MuseScore remains open source, so if one of the users who is affected by this cares to try their hand at this themselves, they would be most welcome to! The more users there are that to depend on MIDI output, the more likely that is to happen sooner rather than later.

Reply

• Jun 3, 2024 - 08:59

Thank you for the information. I did not know that JoJo Schmitz had developped this fork.
I had a look to it; unfortunately, I am not developping using github tools, so I am not familiar with terms like artifact and other elements required to install MS3.7, but I read the installation procedure and eventually succeeded in getting the package and Yessss, it opens M4 mscz files.
Thank you Jojo !

Reply

• Jun 2, 2024 - 02:38

madbrain and Marc Sabatella: I think it was me, and my comment that I could think of cars that are not intended to drive on roads. madbrain, I suggest you actually address individuals in a thread of this length, so that such misunderstandings are (at least) less likely to happen.

madbrain: I disagree, strongly. There is no way that MIDI control can be considered "crucial" to this application. "Highly desirable"? Absolutely. "Disappointing that it's no longer there"? Certainly. But "crucial". Not a chance.

And I don't know what "trolling" or "gaslighting" you think I'm doing. I suspect you may not know the definition of those words?

I am saying, quite seriously, that while desirable, even extremely desirable, it is not important to the core of the software. wcobb suggested that the software was "poor" or "inadequate". I was disagreeing and I still do.

Reply

• Jun 2, 2024 - 11:44

@TheHutch,

Car analogies really don't apply. There is no need for condescension. I understand the meaning of those words, and I stand by them. This is a thread about problems with the MIDI support. You may not care personally about the bugs getting fixed, but others do, and you aren't helping fix them.

MuseScore started as a fork of a MIDI sequencer. It had MIDI support even before reaching 1.0 status. If the developers intended to drop support for MIDI, they would have done so, deleted the code, and announced it. That's not what happened. Instead, they kept the MIDI code, and shipped it with bugs that may or may not have been known at the time. All we are asking is for that code to work at least at the same level that the previous 3.x versions did.

Reply

• Jun 2, 2024 - 17:24

Apparently wcobb (who appears to agree with you) felt that a car analogy really does apply. It's not a great analogy (which is what I was pointing out). but it's not horrible.

No matter where MuS started, what it is today makes MIDI non-central to what it does. I agree that it's nice, but I suspect that the user base for which it is "crucial" for what they do is a TINY fraction of the total users.

It's my understanding that, in the creation of MuS 4, including MIDI was difficult enough that they felt it okay to leave it out. Perhaps it will be included later? I don't know; I'm not in on those decisions.

Finally, again you are accusing me of "condescension", just as you accused me of "trolling" and "gaslighting". I'm not being the one, as I was not being the others. I'm merely explaining my take on this issue. And it's an opinion held by (I suspect) the vast majority of MuS users. The fact that it's important to you doesn't make it important per se.

Reply

• Jun 2, 2024 - 13:36

@TheHutch: You are right: it is not actually crucial for the software. It is a deal breaker for some users and a regression compared with MS3 but I'm sure that MS4 will get there, possibly by the time that it becomes MS5, at which time I expect it to be superior to MS3 in all functionality.

Reply

• Apr 9, 2024 - 04:58

Very same thing happened to me! Doesn't recognize a device but did when I originally downloaded Musescore.
I also lost the ability to use Garageband... informs me that there is an e-licenser problem. Even with a native Mac app like Garage Band!

Reply

• May 29, 2024 - 06:01

It also doesn't work with any of my MIDI devices.
- Munt (MT-32 emulator)
- VirtualMIDISynth (another soft synth)
- Fantom XR (hardware synth)
- VK-88 organ, attached to a UM-880

Not a single byte of MIDI is output to any of these devices.
Also, even if this was fixed, how would I prevent Musescore Studio from outputting its own sounds at the same time ?

I tried the mixer, and putting the volume at zero. Some notes still play on the audio device ! There really should be an option to select "none" for the output audio device if one wants to use MIDI exclusively.

I also ran into the issue where everything plays at about 8 times the speed. Ridiculous bug.
https://musescore.org/en/comment/reply/node/337552/comment_forum#page-t…

It is just bonkers to me that so little testing could be done, or the decision was made to still ship with these major problems. Either way, it's not looking good for Musescore 4.3.0 . Seems like this release was rushed. Definitely not baked properly. Last one I had installed was 3.51 . I'll give 3.6. a chance.

Edit: I tried 3.6.0 and 3.6.2 . MIDI playback is also completely broken for me in both of them. It appears it must have been a long-standing bug. I just don't use MuseScore that often and had not run into it.
However, 3.5.2 is fine with MIDI . and the last one that works for me, in terms of being able to on my MIDI instruments at a sane tempo.

Reply

• May 29, 2024 - 22:44

Hmm, MIDI output definitely works for others, so it could be something unique to your configuration. I recommend starting a new thread and giving more system details.

Keep in mind MuseScore is open source, and testing, like development, is a shared responsibility of the entire community. If no other with the same type of configuration as you chose to take part in the extensive testing period for MuseScore 4, that's unfortunate indeed. But it seems that those who did participate did not encounter this issue. There are other issues with MIDI output, including differences in how velocity and continuous controller messages are sent compared to how some MIDI devices expect them, also a metronome channel that you'd need to disable in your device. But overall as far as I have heard it does work. So again if you're not getting output, best to start a new thread with more system details so other community members can help investigate, and if it turns out there is a reproducible bug, help you submit an issue to Guhtb so the developers can fix it.

Reply

• May 30, 2024 - 00:59

It’s definitely possible to get that information if you are familiar with both the MuseScore source code and with the tools provided by GitHub. I know more than a little about MuseScore source, but next to nothing about the I/O code in particular. And I know only enough about GitHub to do really basic things, not look for all differences in entire folders which is what, I think, this would entail. But anyhow, it definitely should be possible.

Again, though, since MIDi output does work in general, it’s still worth starting that new thread to find out what specific aspect of your system is incompatible with whatever change was made on the MuseScore side. Could be something about your specific MIDI driver, your specific patch version of your OS, or some external device in your chain, etc.

Either approach could lead to discovering the root cause of the issue, but investigating it from the perspective of studying the unique qualities of your system could well get to the answer faster, and won’t require anyone with additional skills to do the source diff. It’s something where you could most of the legwork on yourself rather than having to rely on someone else who may or may not actually be available to help.

Reply

• May 30, 2024 - 10:01

Unfortunately, I'm not familiar with the MuseScore source code. I use github mostly to read and report issues, and sometimes to look at the code as well. But I haven't had to do diffs. Chatgpt told me to use the compare button, but I couldn't find it on github. It was however able to construct a diff URL.

https://github.com/musescore/MuseScore/compare/v3.5.2...v3.6

I haven't had a chance to look at it yet. I need to sleep soon. It would help to know which folder(s) the MIDI code is concentrated in, if it is.

I agree it's likely a bug specific to my configuration. I'm just not seeing it in other MIDI applications.
I will list more details about the config in the thread before I forget them, and then sleep.

Reply

• May 30, 2024 - 10:49

Turns out there isn't really a better place in this thread.

Yesterday, I installed MuseScore 3.5.2 and it played MIDI fine. Upgraded to 3.6.0, and it stopped working.
4.3 also failed to play MIDI. This was regardless of which MIDI device I selected.
The symptoms when it stopped working was no audio from soft synths, and the MIDI activity light remaining off on th hardware MIDI devices.

I'm using Windows 10 x64 with current updates.
There is the following MIDI hardware/software :
- 8-in/8-out Roland USB MIDI interface, model UM-880
- Roland VK-88 organ, hooked up via MIDI to the UM-880
- Roland Fantom XR sound module, hooked up via USB
- Munt, Roland MT-32 software emulator
- Coolsoft VirtualMIDI synth, a soft synth
- Coolsoft MIDI mapper

For audio, I use an ECHO Audiofire 8a Firewire. It has 8 analog inputs , and 8 outputs, and a sophisticated audio mixing/routing app. The Fantom XR and VK-88 organ occupy 4 of these inputs. Can't remember off the top of my head what the other 4 are used for, but they are all in use, as are all 8 outputs.

Earlier today, I uninstalled Munt, VirtualMIDISynth, MIDI Mapper, and both versions of Musescore.
After reboot, I reinstalled both 3.5.2 and 4.0 .

I opened 3.5.2 and loaded this score :

https://musescore.com/user/35237410/scores/6189680

MIDI output was configured as "MMSystem.FANTOM-X" .

I then hit play. I heard the sound of two not quite simultaneous harpsichords. One was coming from the Fantom XR. The other from MuseScore. The MIDI light on the Fantom-XR was blinking feverishly. The VU meters in the Audiofire mixer app were showing signal from inputs 3 and 4 as well. So far, so good.

I then closed 3.5.2 and opened 4.3.1 I loaded the same s Icore. I set the MIDI output to "FANTOM-X".
I then hit play. I then heard the sound of one piano. But it was coming from the scrappy speakers of my LG 32UD59-B center monitor, not from the Audiofire's superb DAC. So, I changed the audio device to the Audiofire. And speed picked up by about a factor of 10x. That part is already documented in another thread/bug.
I then turned down the volume for the Musescore built-in sounds to 0 using its mixer.
I then heard a very faint piano sound coming from the speakers that are hooked up to the Audiofire (with a receiver in between). The mixer app shows signal from the Roland between -58 dB and -60 dB.
The MIDI activity light on the Roland is blinking, though perhaps not as fast as with MuseScore 3.5.2 .
The output level knob on the Fantom XR is turned all the way to the max.
I would qualify this as MIDI not working, though it is different from not working at all.
There were other problems playing that score that were not device/MIDI related :

1) 4.3.1 plays piano instead of harpsichord on 3.5.2 . I might have used custom sounds in my original score. However, it should be possible to fallback to a default harpsichord patch, rather than piano. I seem to recall this was a pre-existing issue in the past
2) the trills are all wrong when playing the score on 4.3.1 . For example, the start of measure 5. They are much slower on 4.3.1 than the same measure on 3.5.2. Perhaps half the speed. Neither is actually the result I want to achieve - I wanted 6 quick notes. The score plays 4 quick notes on 3.5.2, and 4 long notes on 4.3.1. .
A recording on my acoustic harpsichord is at https://www.youtube.com/watch?v=YeDTV8YvwGw .
Correct trills can be heard around https://www.youtube.com/watch?v=YeDTV8YvwGw&t=25 .
I would love to know how to fix them in my score. They were only approximations in 3.5.2, but in 4.3.1 they render the score useless. I wish the program had more intimate knowledge of ornaments, so that one could turn them on or off at will. It is typical baroque practice to play the repeats differently, and varying ornaments is a common way, sometimes omitting them. With repeats, there should be a way to automate that. Of course, there isn't going to be any notation change. This would affect playback only. It would be very useful to turn them off when the program plays the ornaments incorrectly, as MuseScore 4.3.1 does.

There is still a crasher bug I have to describe, but it's not happening anymore right now after the software synths and MIDI mapper were uninstalled. Guess I'll try to reproduce tomorrow.

So much for sleep.

Reply

• May 30, 2024 - 17:28

@madbrain

> I then heard a very faint piano sound coming from the speakers
> that are hooked up to the Audiofire (with a receiver in between).
> The mixer app shows signal from the Roland between -58 dB and -60 dB.
> The MIDI activity light on the Roland is blinking,
> though perhaps not as fast as with MuseScore 3.5.2 .

That is because all MIDI note on/off velocity set to 1 generated by MuseScore4.
Several people has been confirmed and pointed out that problem in this topic.
And, the issue about it on GitHub is here.

"Midi output sets velocity to 1 for all notes"
https://github.com/musescore/MuseScore/issues/22354

Reply

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