MuseSounds in terms of dynamics.

• Dec 30, 2022 - 18:14

Let's just talk a little bit about the weakness of MuseSounds on this wise: dynamics. I am finishing up a string quartet. What I have experienced and analyzed with MuseSounds is that it is not necessarily reading the music consistently dynamic-wise as well as MS Basic, especially for soft dynamics in strings (most likely the lead violin) and the melody that is to be heard. Take this unfinished score for instance. It got to a point where I had to insert all these invisible "p" dynamics or even "mp" in the violin just to make it sound consistent with the rest of the musical phrase, otherwise it would eventually be almost inaudible and undefined. Also compare measures 4 and 22. The clear difference is that there is a consistent dynamic change on measure 22 when it decresc. to "mp" but not so much on measure 4 when it decresc. to "p" and it didn't even decresc. at all until the second note of that measure and then the "p" dynamic is initially nearly silent.
Also, MuseSounds viola in the playback, whenever it releases a loud note, tends to glitch every time, even in a measure of rest. Hard to explain that one better.

It's the little things. Sometimes annoying. Could you imagine how bad MS Basic would have truly been if the dynamics (especially soft) weren't straightforward and consistent for at least four measures?

Dynamics are, I think, the biggest weakness in MuseSounds thus far, but I am excited for future updates, because this is not a finished work and we soon won't have to talk about this. And also, I contend that the playback is just another little part of the MuseScore 4 experience. Realistic computer sounds awe me.


Comments

Nadine, I agree with everything you say. The dynamics in 4.0 drive me up the wall.

However, I have discovered that the new MuseFX that I downloaded through Muse-Hub can make a significant difference in how 4.0 handles the playback dynamics. It is far from perfect, but still a huge improvement over the default 4.0 playback. Note, however, that the Linux version of 4.0 is cripple-ware, so you have to be using Windows, and, I suppose, or Mac to take advantage of the MuseFX.

I have not discovered any documentation for MuseFX, but through experimentation, I have found that I can even out the dynamics of 4.0 somewhat using several of the FX plugins. You can apply multiple FX to each instrument and also to the Master channel.

In particular, I have found the "SimpleEQ" plugin to be quite helpful. Although the MuseFX do not offer the control that most FX plugins offer, their simplicity makes it fairly easy to experiment with different settings. Each plugin has multiple presets and a single control knob to adjust the amount of FX that is being applied to your playback. You can turn the effect on and off from the mixer panel so it is easy to compare the difference between plain 4.0 playback and the FX.

Incidentally, the MuseFX also work in other software packages. They work in DAWs like Reaper, Audacity, even Dorico! Give them a try.

In reply to by Jim Ivy

@Jim Ivy: You said, "...the new MuseFX that I downloaded through Muse-Hub ... however, the Linux version of 4.0 is cripple-ware, so you have to be using Windows, and, I suppose, or Mac to take advantage of the MuseFX."

That's why I see nothing about MuseFX when running MuseHub on my Ubuntu Linux??

And "...[MuseFX] works in DAWs like Reaper..."?? Perchance, have you tried this??

In reply to by Are Jayem

Yes. Musescore Group seems to be more interested in catering to Windows and MAC users. Of course that is not unusual, but Musescore would not exist had it not been for Werner Schweer and the Linux users who originally created it.

I installed Windows in a separate boot partition specifically to be able to evaluate that version of Musescore. So I have been jumping back and forth between the Linux version and the Windows version. In general, the Linux version is more stable and does not have as many glitches as the Windows version. But, as you noted, the Linux version does not allow using VSTi instruments or the FX. However, if you install the fx on Windows, you can copy them to Linux and they work just fine in everything EXCEPT Musescore!!!!

The MuseFX are simple to use because they depend upon presets. The only controls are for preset selection and the amount of effect that will be supplied to the track. This is inline with Muse Groups stated goal of making music creation tools available to everyone no matter how little they know and understand about music.

In some ways this is a laudable goal. But it is also an indication Muse Group is more interested in the "hobby" market than they are in the "musician" market. It particularly does not bode well for those of us who prefer the stability and utility of Linux over the marketing platform called "Windows".

Musescore seems to be committed to their stupid Muse-Hub process. That is a blatant marketing move on their part and the major reason I am actively searching for an alternative to Musescore that follows the principles that are embodied in the Linux gestalt.

In reply to by Jim Ivy

If you find the alternative to MS, please contact me through the .COM profile Are Jayem. (Hey, triple-WOW on your 'Triptychos' score...) For all of 2022, I've nearly doubled my compositional output, all done on a "thin-client" HP system. My main studio system is an iMac from 2013, which I'm concerned may have OSX revision challenges to run 4.x of MS. Sad but the fact of life the 800-pound gorilla called Windows so-fully impacts developers. But kudos to MS for endeavoring to be cross-platform. And into mobile devices, I sense. Another fact is, us end-users can swell protests to the Muse Group, but they will go their own way, I'd think. Look at us venting so early in 2023.

In reply to by Jim Ivy

Thank you. I find this post amusing. Yes, Linux is light weight and free. And used worldwide in great numbers. I get it. But one of its founding principals is that the user must be computer literate. Most of us (me included ) are not . I tried Ubuntu for several months and found it to be new user un-friendly. I had to find different versions of everything I was used to using. I finally gave up because I spent more time fiddling with the OS than actually using software (much of which was not as good) on it.
So, indeed an OS it a tool. So lets look at a common tool. One version is by Linux and the other by most any other OS.
That tool is a hammer.
While Linux is free, the device it is installed on is definitely not free. So the Linux hammer comes in pieces. Some pieces we may have already had. Other pieces we had to buy. So we assemble our hammer. We may or may not know what we want to use this hammer for. But we wouldn't want to use a hammer with a wooden head to drive nails. So we just happen to find the metal head and put it on the handle. It doesn't fit. We are hammer literate so after a while we figure it out. And because we are hammer literate, we also figure out how to make the head stay on the handle so it doesn't fly off as we swing it. And also because we are hammer literate, we find and install some kind of grip on the handle ,(or at least shape the handle) so it doesn't fly out of our hand as we swing it. Perhaps gloves and eye protection would be nice. But we are hammer literate, so those aren't important at all. Now we can get to work.

Any other OS hammer. Several are on display at the hardware store. Ready to go. And we can get to work.

The Linux gestalt certainly is great for people who are interested in working that way. But it just isn't for everyone. I think most people just want to turn of their computers and get to work

In reply to by Jim Ivy

Extremely well written. And 1,000% correct.
Frustrating beyond belief.

But it is also an indication Muse Group is more interested in the "hobby" market than they are in the "musician" market.

Musescore seems to be committed to their stupid Muse-Hub process. That is a blatant marketing move on their part

Consecutive bullseyes. Bravo, sir.

This doesn't help you much, but as I understand it, part of the goal of muse sounds will be to interpret scores along the line of NotePerformer. Something that I'm not interested in. This seems to affect solo sounds and not section sounds as yet. It's also not working well, obviously. Sometimes the oboe will be playing at a P dynamic just fine. Then two measures before an mf it will start to get louder all on its own.
I notice in your score the Ist violin plays a bow direction change in measure 7. Making two quarter notes. When I change the voice to Violin 2 solo, it plays a half note. Makes you wonder what else is going on.
Who knows when or if any of these things will be fixed/completed.
An ongoing problem with any notation software is the battle between what you want your score to sound like and what you have to do to get there. And what works in one program won't work in another.

In reply to by bobjp

Well, you have put your finger on one of the big problems I have with 4.0. It makes decisions on its own that may run counter to what the composer/arranger wants. Part of the performer's role is to look for expressive possibilities. Even something as simple as an ascending major scale has many different expressive possibilities. I really do not want a robot deciding to crescendo in places where I want the dynamics to stay even, or vice versa. Even if the robot makes the right decision 90% of the time, there will still be places where what I want will not be what the robot plays. I am not totally opposed to some sort of artificial musical intelligence, but I want to have the option of overriding it when it does not fit what I am trying to express. At the very least I want to know explicitly what expressive rules 4.0 is following when it changes something that is written.

In reply to by Jim Ivy

People love NotePerformer. It's supposed to render scores in a more natural, musical way. Phrasing and what not. It uses its own sounds which I don't think are as good. And, as you say, what if I disagree with how it plays something?

In reply to by bobjp

I believe that the musicians job is to look for expressive possibilities, to choose from those possibilities, and then present them to the audience. Depending on you age, you may or may not remember Dick and Jane. Dick and Jane were the main characters in the books that were used to teach children to read in the 1940s and 50s. Dick and Jane communicated through simple subject/verb/object sentences.

When I was teaching music, even with beginning students, I wanted them to understand "expression" in music. So I would write three words on the black board, something like, "Dick saw Jane", and ask each of my students to speak the words but to make them mean something different from the meaning the other students had given them. The point was to illustrate the difference in meaning that vocal inflection, pauses, tone of voice, etc. can make. Then I would have each student play something simple; perhaps the first five notes of a scale, but they had to make those notes "mean" something different from anything the other students had played.

So that is my problem with programmed expression. Yes, there are certain general principles; we generally play louder as the pitch rises and softer as the pitch descends, but musical expression lies in the exception, not the rule.

In reply to by Jojo-Schmitz

Yes, punctuation is a very important part of written communication. Still, when I read Shakespeare, I don't always get the same sense from it as when I listen to Laurence Olivier or Richard Burton speaking the same lines. We may be reading the same words and the same punctuation, but they bring insights that never occur to me.

That is the way that music is. It does not matter how carefully the score is notated, the performer brings insight that perhaps even goes beyond the composer's original intent.

I'm not an expert of MuseScore (despite having been using now and then during the years), but I'm facing the same problems with dynamics.

Premise: I'm working on a score that I've entered and edited in Dorico with NotePerformer. The XML transferred generally well, and the dynamics are the ones that were in the original score.

However, using the basic MuseSounds, the same dynamic marks don't mean always the same. 'ff' may be just that, or a booming explosion. 'p' may be that, or something inaudible.

Also, when going from a 'p' to an 'f' dynamic, I usually hear the loudness going up a little before the interested note, as if the dynamic change is anticipated by a few MIDI ticks.

Just wondering if I must attribute these issues to a bad conversion of the XML file coming from Dorico, or if the behavior of the dynamics is really different between the two programs, and one is expected to retouch them after converting.

Paolo

In reply to by PaoloT2

In my experience, no two notation programs, just like no two musicians, will play forte at the same volume. When I work on something in MU4 that I started in Sibelius, I mostly have to redo the dynamics.
I also sometimes get a pre dynamic volume increase. When that happens, I put a hairpin on the last note before the dynamic. Should I have to. No. But I need to get on with working on the project.

In reply to by PaoloT2

@PaoloT2 wrote > ... using the basic MuseSounds, the same dynamic marks doesn't mean always the same. 'ff' may be just that, or a booming explosion. 'p' may be that, or something inaudible.

Also, when going from a 'p' to an 'f' dynamic, I usually hear the loudness going up a little before the interested note, as if the dynamic change is anticipated by a few MIDI ticks.

I've encountered these exact issues in MuseScore 4, but haven't tested to observe the behavior of such matters in MS4.4.1 ...

But I did make this Github report recently regarding velocities and MuseSounds in MS4.4.1. The library used was Acoustic Nylon and the results are needlessly erratic.

Just wondering if I must attribute these issues to a bad conversion of the XML file coming from Dorico

Why not reconstruct an example phrase directly in MuseScore Studio and see what you get? I'll be interested to know!

In reply to by PaoloT2

Slightly before the onset of the C note it sounds like the tail of the B jumps to forte level output, or something close to forte. I hear the same effect if I set the opening dynamic of mp. The effect is less obvious if the opening dynamic is mf

And if I set the tempo to 40 BPM it's sounds like the transition takes approximately the same length of time.

Do you know how to report an Issue Report to Github?

In reply to by scorster

Yes, I can report it on GitHub. I'll see if there is already some thread about it.

If you have a clearer idea of what's happening (as I said, I'm not an expert of MuseScore), maybe you can report it in a more detailed way (as you just did here above)?

Paolo

In reply to by PaoloT2

@PaoloT2 wrote >

> Yes, I can report it on GitHub. I'll see if there is already some thread about it.

Great!

> If you have a clearer idea of what's happening (as I said, I'm not an expert of MuseScore), maybe you can report it in a more detailed way (as you just did here above)?

If you pot to Github, feel free to combine my wording with yours. Also:

  • please paste the Github link here. I'll comment at the link if needed.
  • and post a link to this thread at Github

Sorry to say, I rarely have much clarity regarding the numerous MuseSounds anomalies. They are pretty impressive libraries that leave me waiting for the day when I can have a largely glitchless time with them.

Thanks for this!

I´ve also noticed some dynamics issues, f.e.: in a String Orchestra piece I´m working on right now, a mf sounds sometimes like a ff and a mp like a pp. And then, in an other section it sounds just right. So there is also an inconsistency there. But I´m not worried about that, as I´m making a final soundfile in my DAW. But of course, it would be great if this woud be possible to achieve allready inside the notation software.

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