Default and max playback volume much too *loud*

• Oct 26, 2014 - 16:27
S4 - Minor

Currently v1.3 provides a volume slider on the playback panel. This is helpful because sometimes the playback sound intensity from multiple instruments (in the aggregate) is much louder than sounds from other software programs on the computer, and so a playback volume control allows users to turn down the volume of MS alone instead of changing the existing speaker sound levels on the computer.

However, the current implementation is not doing that. On my v1.3 (Win7/x64), installed last night, the MS volume on a new score is much too low, even with volume in the playback panel set to the max. It forced me to turn up my computer volume, which then caused system sounds (dings, incoming mail, Skype, etc) to be much too loud.

I had no dynamics (mF, F, etc) in my new default score.

I downloaded some other scores, and they played at an appropriate sound level. I noticed that they had dynamics, and that clicking on the opening dynamic showed me a MIDI sound level. But the score that played properly had a dynamic that I didn't recognize (like a scrolled f), and it had a MIDI level of 1 (I couldn't understand how a 1 could produce that much sound).

I added some test dynamics (eg mF), and the playback sound improved a little. But I do not think that should be necessary. There is a volume issue at work here. Perhaps I am doing something wrong, but please keep in mind that:

- it is a fresh installation of MS
- I am a new user (who probably doesn't know enough to screw up volume settings)
- it is a fresh new blank, default (piano) score, with only a few lines of simple notes
- there were no dynamics in the score text
- the playback volume was set at the default (90%?)
- the mixer controls were set at the defaults (I didn't touch them)
- I searched the doc, but could find nothing useful about this situation

For now, I am forced to play MS at a very, very low volume, because the pain of increasing my default computer speaker sound levels is too high. I wish I knew enough about MS code, score, and operations to more clearly pinpoint the technical issue, perhaps by more contrast between my score and other scores, but unfortunately this is the best that I can do at this time.


I am not understanding your report. You say "the current implementation is not doing that. On my v1.3 (Win7/x64), installed last night, the MS volume on a new score is much too low" are you actually talking about a new installation of 1.3, or of a Nightly build of 2.0?

If it's a new installation of 1.3, then probably you had turned upt he Synthesizer volume in your previous installation and you simplky need to do the same in the new one. And in any case, 1.3 is not receiving any further fixes - all current work is directed at the upcoming 2.0 release.

If it is a nightly build of 2.0 you are talking about, please post the score you are having problems with. The fasct that other scores apparently play correctly suggests that somethng might be wrong with your score only. Maybe also post one of the scores that is working correctly for you.

FWIW, whether you add a dynamic marking or not should not be relevant. If you add no dynamics, things should work the same as if the dynamic were "mf".

In answer to your questions,

My new installation was of v1.3, downloaded and installed for the first time.

I don't recognize the Synthesizer volume term; I didn't set it in MS (if that's where it resides), and it is not visible in my Volume / Mixer in the system tray on my Win7 machine.

I agree and understand that 1.x versions get no new development.

The volume issue is a puzzle to me, because I went through the same kind of software/debugging thinking that you describe - was it something that I did? already on my machine or in my machine settings? in my default / simple first score? in a play panel or something? But I came up with nothing, and thought therefore it was worth reporting, given the fresh installation, the simple default score, etc.

For now I'll just live with ("manage") the quiet volume until something changes or 2.0 comes out.

Status (old) needs info closed

The synthesizer is a window within MuseScore - see the Display menu in 1.3. This is where you can control the overall volume of MuseScore. Probably at some point in the past you turned it up, and now you need to do the same for your new installation.

It's ok with me if this issue is closed; I'll just live with it until 2.0 comes out (and hopefully the issue will just go away)

But just to be clear, I would prefer to state the issue closure is definitely not because I messed with the synthesizer volume -- I didn't even know such a thing existed. And this is my first install of MS, so there was no previous version.

Based on your post above, I found the synthesizer window in the Display menu, and the volume was set at 10 (but I'm guessing that's 10% down from max, since the scale reads 0->50, with 0 being the loudest. So even if I had somehow set it too low before (definitely not the case), I could not fix things by turning the synthesizer volume up again. (Here I assume that 10 is the default setting, since the color bar (VU meter look alike?) goes red into some danger zone above 10.

See the attached screenshot.

Attachment Size
2014-10-26_17-02-21.jpg 54.6 KB
Status (old) closed active

I found out that MS 2.0 Beta was available, and installed it. The doc said it was ok to install beside my v1.3 that installed a couple of days ago. I am posting to this thread for the 2.0 version, even though the original thread was for the 1.3 version.

I have set the status to 'active' so it will show up on someone's list.

- Win7/x64 desktop
- fresh v2.0 beta install
- someone else's score (Nicolas Beethoven Moonlight Sonata)
- touching NO volume settings in MS 2.0 AT ALL
- I just loaded the Moonlight score from the 'Connect' panel, and hit the play button

- the MS playback volume is so low I can't even hear it.
- if I crank up my speaker levels, all other apps BLAST out loud sound

- ITunes sounds are normal (MP3s are easily heard)
- System sounds are normal (dings etc are easily heard)
- Windows Live Mail dings are easily heard
- YouTube videos in Internet Explorer are easily heard

I have attached a screenshot showing simultaneous sound settings for MS and my system mixer.

It seems to me that my computer, and all other sound-producing apps are working fine, and that MS (both 1.3 and 2.0) are the exception. I would be grateful if someone might have a solution to this problem. Thank you.

Attachment Size
MS Beta 2.0 Moonlight Sonata.jpg 146.53 KB

And just in case someone argues that my system sound levels are set too low, I have already investigated that. I turned down my physical speakers to the minimum, and lifted my system mixer levels up to 50% midrange, so that ITunes, YouTube, LiveMail, and System Sounds were all easily heard.

See the attached screenshot.

The Moonlight score in MS was still too low to hear (if I put my ear up close to the speaker, I can barely hear the loud sections). Thank you.

Title playback volume too low even with playback volume set to max Default and max playback volume too low

Hey kkkwj, your observation in comment #6 is correct and has been raised before. It's an open issue.

What is the dynamic level of the score you are testing with? If it's i herently qpa quiet score, it is normal that the playback level would be soft. Most other applications deal with "conpressed" audio where are sounds are at close full volume with very little dynamic range, but MuseScore does not - it tries more faithfully to teproduce real dynamics as played by real musicians. So it normal that notes played "pp" would be much quieter than system sounds. If this were not the case, then there would be no way for notes played "mf" to sound louder and then "ff" to sound louder still. You can use the system mixer to turn up the volume for MuseScore if your OS supports it, but you will then find that notes played "ff" are louder than other applications.

That said, there is a *slight* difference in volume levels between 1.3 and 2.0 using their respective default soundfonts, and this is somethi that probably ahould be addressed.

Thanks to Thomas in #8 for saying that the issue is known and open for other people too.

For Marc, the score I'm using (Nicolas Sonata No. 14, Moonlight, Beethoven) has no dynamic symbols at the beginning of the score. Please see the attached image.

Given Thomas' statement, perhaps my issue falls into the scope of the previously known issue (Thomas, is there a title or an issue number for that issue?). So I thought I might investigate further, to see if other scores had the problem (or not).

All these scores are Moonlight Sonata, No 14, from MS Connect:
by Nicolas, no dynamics, has the problem
by Chopinowsky, no dynamics, volume ok
by LucyJoDixon, no dynamics, has the problem

Hopefully identifying the scores might help to debug the problem.

PS. I forgot to include the MS mixer in the screenshot above showing all my sound levels. Of course I had not touched the MS Mixer in the fresh 2.0 Beta install. But here is a screenshot of it anyway. It shows the default settings (vol = about 2:30 oclock, maybe 75%?)

There is no open issue for this as far as I know. It's just something I have seen and read multiple times. I consider it a major UX issue. But I believe it's not an easy fix.

The fact that some scores work, and some don't, on a fresh installation (see my previous postings) gives me some hope that it might be an issue in the score files alone. I find myself wishing for an "XML" diff program that could understand MS score files (at least for the general or high level config setting stuff). Not that I would know what to do with the diffs anyway, since I don't understand MS internals.

The issue may also be caused by the interaction between the score files and the default program settings, of course.

Thomas, could you please say a few words on why you think it's not an easy fix? And what strategy you think might be best for investigating (and hopefully solving) the issue?

Could you attach the specific score you are having problems with?

The one I found uploaded by Nicolas most definitely has a "pp" at the very beginning, and that totally explains why it is quiet - as this piece of music *should* be.

Even if you don't see a visible dynamic marking, it's possible a score has used Note Properties / Insepctor to explicitly set the velocity of individual notes to be quiet. That is why it would be useful if you posted the specific score you are having problems with.

Again, *in general* there is no actual problem. There are a number of reports like yours that turn out on further investigation to be similar misunderstandings. There is also indeed the *slight* overall difference I already mentioned between 1.3 and 2.0 that is, I believe, caused by the difference in recording levels of the default soundfonts (using the same soundfont in both and similar synthesizer volume settings, volume levels are probably the same).

Assuming the slight difference is due to differences in the soundfonts themselves, it would be possible to increase the volume of samples in the new default soundfont (FluidR3), but some amount of work would be required to do it right - preserving (or improving) the relative volume between instruments. Eg, "mf" on a flute has to be quieter than "mf" on a trombone in order for ensembles to sound correctly balanced.

@kkkwj if you could reproduce the issue with one or more specific file, this will certainly help when I report this to lead dev Werner Schweer. Thanks for your help!

I would be pleased to help by clearly identifying the scores with issues, and thought I had already done so in my previous post. Here is the list again, all from the MuseScore Connect panel in v2.0. I had clicked the Nicolas' version at random to start with, and simply searched for Sonata No. 14 to see the rest. So I thought anyone could fetch them there, like I did. If I post them, it means they have to pass through my hands and my computer, and I run the risk of someone saying that "maybe the files were corrupted passing through my hands..." or whatever.

Regarding the pp in the Nicolas version:

I anticipated the "well the score has a pp, so the volume is supposed to be low" argument, so I checked for dynamics (as I always do on this problem). I saw a tiny pp buried in the long (Italian?) phrase. I had typed that observation into my previous posting, but questioned whether it had any effect, because I went to a new score, added a pp dynamic, and the added pp font was different, so it was not clear to me at all that the Nicolas embedded pp was _functional_ in any way, being embedded in the middle of a foreign language phrase "sempre pp e senza sordini" that I could not understand. So I removed the typing that I had done in the posting. (As I type this sentence, I wonder if Google translate could have helped me.)

_In addition_, I loaded some other Sonata 14 scores from other people, and those scores had (1) no dynamics that I could see, (2) embedded in foreign phrases or not, (3) and but still had the volume problem.

So I reasoned that the embedded pp (if it was a functional dynamic at all; that's not clear to me) in the Nicolas score could not be the problem (or be relevant) in itself, since other scores of the same Sonata had the same sound (ie, essentially none) even though they had no dynamics at all.

You can see that it is consistent for me to question the functionality of the Nicolas embedded pp if other similar scores had exactly the same volume issues, without having dynamics at all.


I do want to comment that I feel like I'm on the defensive in this thread. Thomas has been the only person to add a comment that he thinks it's an issue because he's seen other people report it too many other times.

Several other postings feel (at least to me) like they point the finger at me (it's my fault for early installs, my machine settings, my scores, etc.) I spent 25 years of my career running a small software company, and have spent decades reading/participating in forums, so I do make this posting lightly. I know that there are thousands of issues in forums, and especially in open source forums, where all manner of posters can post stupid or uninformed things, and that forum postings (like email) can seem more aggressive than they are meant to be.

Please give me the benefit of the doubt that I did not post this thread on a whim -- it costs us all precious time and effort to deal with the thread alone, let alone the underlying problem. If there is something reasonable you need me to do that I am capable of to help solve this problem, please say so. But please try to accept that this is a real issue. Thank you.

I am sorry that you feel like you are on the defensive, and am equally sorry if my posts have somehow come off as aggressive. This is not my intent. I am simply trying to explain the behavior of the program to you. I really do understand what is going on here, and based on the available evidence thus far, there is *not* an actual problem here aside from what I have already acknowledged - a *slight* decrease in the volume of certain instruments from 1.3 to 2.0 due to the different levels at which their soundfonts were recorded (other instruments are actually louder than 1.3).

Sio please don't take my explanations personally. I ask that you give me the same "benefit of the doubt" that you are asking of me.

I absolutely agree that MuseScore's playing of "Moonlight Sonata" is quieter than one's system sounds. I am simply trying to explain that is intentional and correct - it is not due to some bug, but due to the fact that MuseScore does honor dynamics and velocity settings in a score. MuseScore is simply doing what it is told. The score contains an explicit instruction to play quietly, and MuseScore is honoring it.

I asked you to post the specific score you are having trouble with to remove the possibility of doubt - I thought maybe you were using a different version than the one I am. But from your description above, it is clear we are both talking about this one:

Indeed, the "pp" marking is buried with the larger text, but if you right click it and check its properties in the Inspector, you can quite clearly see it is a real Dynamic marking, with a velocity of 33 (out of a maximum of 127).

So indeed, this piece *is* specified with a very quiet dynamic marking, and MuseScore is simply doing what it is told here. 33 is basically 1/4 of 127, and we hear things somewhat logarithmically so it may well *sound* even quieter than that. So of course it comes out much quieter than other sounds on your system, that are usually played full volume.

If you delete the dynamic marking from the "Moonlight Sonata" score, the default before "mf", which corresponds to MIDI velocity 80. So your score will still be playing at only roughly 2/3 volume (and may sound less because of the logarithmic nature of hearing). But I think you will agree it sounds more comparable to other sounds on your system at that point.

Again, if you have some *other* score that is *not* set to be be playted quietly either via dynamic markings or direct note velocity settings, feel free to share it. Maybe there really is some issue with some scores that is *not* explained by dynamics and velocity settings. But for now, this issue can only be considered "by design". MuseScore is doing what it is told. Anything else *would* be a bug.

Thank you for your post. It helps to read that you do not intend to be aggressive.

I do not think the pp is relevant to this discussion, because (1) other scores without pp have the problem, and (2) even if pp was the issue, of course it should not drive the volume on my machine towards zero. I should still be able to hear the music without effort. Neither do I believe that "MS is doing what it is supposed to do by design -- how could it, when multiple scores have essentially zero volume level?

Nonetheless, because the pp is the central topic of the thread right now, I'll explain my view once again. Focusing on the pp Nicolas score - of course the first thing I did before posting was to look for a dynamic, and to right click on the pp (it was a slightly different font on my screen in MS v2.0). But nowhere could I see a MIDI level value in what popped up "Edit element" showed nothing relevant to MIDI levels that I could see. It was all simply text.

Marc keeps talking about an "Inspector", but I don't know what he means. I right clicked again, and thought maybe he was talking about the "Debugger" option on the right-click menu. I have included a screenshot of the debugger dialog window from the Nicolas score, with some red arrows. The debugger shows the word Dynamic in the left XML? structure, and the XML? SGML? tagged text itself, but nowhere do I see the MIDI 33 velocity that he mentions. I do see a 0 in the bottom right Dynamic box, but it is not obvious to me that the 0 has anything to do with MIDI velocity.

Please see the Debugger screenshot (hopefully Debugger = Inspector?)

And once again, here is the previous list of scores that I used for my investigation:

All these scores are Moonlight Sonata, No 14, from MS Connect, using MS v2.0 Beta:
by Nicolas, no dynamics (has an embedded pp), has the problem
by Chopinowsky, no dynamics, volume ok
by LucyJoDixon, no dynamics, has the problem

Attachment Size
MS Beta 2.0 Moonlight Sonata pp.jpg 138.82 KB

Actually, the Debugger and the Inspector are different. To display the Inspector, you can go to the top menu View and select Inspector. You can also press F8.

Once the Inspector is displayed, you can click on an element (like dynamics) in your score and its property will be displayed in the Inspector.

OK, let's first establish: do you now agree Nicolas' version does *not* demonstrate a problem? It has a "pp". Doesn't matter if the text happens to be buried in a larger string - MuseScore doesn't go hunting around in strings. It is either valid dynamic marking or it is not. And if it is, MuseScore uses the velocity value set for that dynamic marking, not anything about the text. The text could say "as loud as possible", or "think about pink bunnies" for all MuseScore cares. All it knows is the velocity of that dynamic marking is set to 33, which is indeed very quiet. So this score behaves exactly as expected. Mostly likely, Nicolas used a "pp" from the palette then edited the text.

The second score you mention - from Chopinowsky - you also say has no problem. Great.

So *if* there is a problem, it exists *only* with the third score you mention - from LucyJoDixon. I did a search online, and found this one:

Well, this one also has a very plainly visible "pp" right at the very beginning. And again, if I view it in the Inspector (which hopefully now you have found) you can see it really is a dynamic marking, and is indeed set to 33.

So again, there is no problem with any of these scores.

Once more, if you find a score where you think the playback (or anything else) is wrong, please post the score itself, or at least a link to it, so we can quickly be sure we are talking about the exact the same score. Maybe you found a different version of this score also posted by LucyJoDixon. But the one I just posted the link is very clearly meant to play back quietly, and it does.

Thanks to jpfleury for pointing me to the inspector. I was able to see the 33 velocity that Marc was talking about. Thank you; no doubt that will help me to provide better info in the future.

On my machine:
- the Nicolas score _does_ have the problem (it also has a pp).
- the LucyJoDixon score _does_ have the problem (it also has a pp).
- the Chopinowsky score does NOT have the problem (it has no pp).
In this score, the Inspector says the velocity is 0 and of type offset (presumably from the default of 80?)

Marc also makes a valid point that we hear logarithmically, so pp might sound a LOT less than the MIDI levels might indicate. (Eg. at velocity 33, 1/4 of 127, it might sound like 1/8 instead (which I think it does).

Following more of Marc's suggestions, and using the Inspector, I increased the original 33 up to 80, and finally could hear some sound out of both "problematic" scores. I could delete the pp (as Marc suggested), thus invoking the default mf = 80, and could hear the score.


With the help of the nice Inspector and contributors to this thread, it seems clear to me now that Marc is correct on his main points about:

(1) MS generates, by design, a huge dynamic range,
(2) that pp at 33 is far below normal system / YouTube / ITunes sound levels,
(3) pp is so far below normal that you can't hear it (with logarithmic ears), and
(4) that my issue is one of understanding, not one of machine, MS software, or score.

Further, I cannot reproduce the original issue (no sound on a new, no-dynamics score in v1.3) that prompted me to start this thread (see my original entry on this thread). Neither can I recreate the low volume experience on a new score (no dynamics) in v2.0. And although I cited the Moonlight scores as evidence to support my original issue, I can see now that they are behaving technically correctly, so they do not support my original case either.

Since I can't come up with reasonable (and relevant to v2.0) evidence to support my original issue (with v1.3) I must abandon it.

Finally, I gently suggest:

(1) that it might be useful to write up a section for the 2.0 manual on understanding sound levels in MS, so people don't have to go through this exercise again, (even with Thomas supporting them :-)).

I volunteer to write up the section based on the knowledge I have gained from this thread. I can write up the text and images in a Word doc, if someone else can cut/paste it into the manual in an approved way. (I would prefer not to have to learn all that other stuff at this point in time.) Give me an email address or process to follow, and I'll send in my work.

(2) that it might be useful to reconsider the default sound levels bound to pp (and friends) so that they might actually be heard on computer speakers that are set comfortably for other system sounds, ITunes, YouTube, etc.

I cannot see the point of having pp sounds so low that people can't hear them on their computers. As Marc said earlier, "keep in mind the main purpose of MS is to be a notation program that can generate written scores for real musicians...", and I cannot see how super-wide dynamic ranges (that effectively interfere with normal computer sound level settings) promote that goal.

If super-wide dynamic range is _really_ worth the price of having future people go through the process that we just went through, then maybe consider an MS option that allows custom binding of dynamics to different MIDI levels, so people can just choose a mapping that meets their needs. The default, of course, should be something they can hear on their normal work computers.

Sorry once again for all the thrashing, but at least we have a link we can point to in the future that will hopefully shorten future discussions.

This issue can be closed.

Just for completeness, here are the MIDI levels that the inspector just showed me for v2.0:

ppp 16
pp 33
p 49
mp 64
mf 80
f 96
ff 99
fff 99

Obviously, this thread shows that I do not have a deep understanding of why these MIDI levels are set where they are. I have only used the default piano and violin sounds; other instruments might be different.

But it seems odd to me that the default level mf=80 is so close to the max dynamic level ff = fff = 99 (only 99-80 = 19 ticks difference out of 127 max), and yet so far away from the minimum dynamic level ppp=16 (80 - 16 = 64 ticks).

Quite a bit of asymmetry in those settings. So much asymmetry so that you can't even hear pp = 33 (47 ticks away from mf=80) on your computer with normal sound settings. And ppp=16 (64 ticks away) would be even farther below audible level.

Maximum dynamic levels ff = fff = 99 are a bit louder than mf = 80 to my ears (only 19 ticks away), but not by much. I would have intuitively expected a FAR louder sound at fff, comparable (by symmetry) to the quietness of pp or ppp.

OK, glad we've got that much straightened out!

I take it you are using the beta or another ealrier build of MsueScore - that would explain the 99's you are seeing. That's a bug in the Inspector (not in the actual playback) that was fixed a few weeks ago. The actual values are 112 for "ff" and 126 for "fff". So they aren't as oddly skewed as it might appear.

Anyhow, the thing is, whether or not you can hear the playback at velocity 33 is entirely dependent on your system volume, speakers, etc. I don't think MuseScore can be in the business of aritifically constricting dynamic range just because some people might have their syste, volume turned down or might have less powerful spekaers. Real music has a gigantic dynamic range, and MuseScore is trying to be as accurate as it can be. Is there room for improvement? Probably. but I don't think you can make determinations based solely on how things sound through computer speakers with the system volume turned way down. Even at "fff", the volume would be way, way below what any real instrument played "fff" would sound like. I think the goal should be, if you have your sound levels set so that "mf" sounds similar to a real instrument instrument at "mf", then "ff" and "pp" should scale accordingly. So that if you are setting up for realistic playback it will actually sound realistic.

That said, I wouldn't be completely opposed to some sort of compressed mode that deliberately and artificially compresses the dynamic range, for use when playing back when system volume is turned down or when playing through weak speakers. But it would be impossible for MuseScore to determine what kind of sound system it is playing through, so this option would have to be set manually.

Just so that the OP understands, and this should be included in the handbook if it's not, MS has several "volume controls" that affect the playback level:

1. The dynamic marked in the score, or the default of mf (80)
2. The Play Panel volume slider
3. The Mixer volume controls
4. Manual note adjustments in the Inspector

And then add in the Windows volume mixer controls, and this makes for a fun time to sort out volume issues. Generally, when I want to play a score in MS, I find myself turning up my stereo quite loud because of soft playback. I generally leave the Windows volume controls up full, so that's never the problem, but there 3 other places to check/change the playback volume level.

I'm glad to hear that you would not oppose the idea of an option that would use "compressed sound levels" to make the default MIDI levels sound "workable" on a normal computer (without changing any of the normal levels inside or outside of MS). A default mode like that would prevent this whole "why can't I hear the score" process in the future, I think.

Provide the advanced "wide dynamics" MIDI level mapping 16-126 option for advanced users who have great speakers and who know what they're doing. But for most newbies and normal users, I'm thinking a "compressed mapping" (as you call it) is a far more workable and realistic solution.

The thing is, everyone has their own idea of "normal". I think my computers are pretty normal too, and I can hear "pp" just fine. It's quiet, as should be expected, but very plainly audible. This is with the built in speakers on three different laptops. The difference, I guess, is that I "normally" keep my system volume at full or close to it. If the sounds from the OS are loud and bother me, I turn *those* down using the OS system mixer - I don't use the master master volume for this (and that is what I usually do on my Windows computers).

And really, the sound of MuseScore playing "pp" is not really obviously softer than the sound of any other audio program playing a recording of a piece that was itself played "pp", assuming that recording is not itself artifically compressed (as most YouTube videos probably are).

So I think it would be a mistake to add *much* compression by default - it would come off as a bug to anyone who has their speakers turned up to a "normal" (there's that word again) volume. But maybe a *little* compression would not be too harmful.

BTW, when I say too much compression would look like a bug to those of us with what *we* think of "normal" computers and speaker settings, here is what I mean.

Say we made it so "pp" was the equivalent of 80 - what you are implying would be the expected volume. Now you can hear it, but on my computer, that isn't going to sound like "pp" *at all*. It's going to sound like "mf". Same without anyone else with speakers set more like mine than like yours. And again, I am talking about ordinary laptop speakers on three different laptops - nothing fancy. So anyone with speakers set up like mine would instantly complain, "I put a pp dynamic marking in my score but it is not playing back quietly at all", and they'd be absolutely right to call that a bug.

Furthermore, consider - it is only 32 steps from 80 to 112 (the velocity value for "ff"). Yet we need to cover 5 dynamic markings in that range - from pp to p, mp, mf, f, and then ff. That means the difference between any two dynamics would be only about 6 steps. For example, "mf" might be 96, "f" 104. Try setting those velocities in the Inspector and tell me how much difference you hear. It will be barely detectable. So not only will people be complaining that "pp" doesn't work, they'll be complaining the dynamic markings don't seem to work period, because only extreme changes (eg, from "f" to "p") will be detectable at all.

That is why we really cannot be in the business of assuming people will have their system volume turned down low and therefore applying lots of compression by default. It will pretty much destroy the effectiveness of playback of dynamics.

Only question is whether a *slight* compression - one that still allowed us to hear a clear differences in dynamics, and in which "pp" was still perceived as "very quiet" as it should be, but maybe raised the level of "pp" a little - would make any difference in your case.

Those are all reasonable points about defining "normal."

But having spent a lifetime with computers and computer users of all kinds, I would easily say that anyone (eg. you) who has the understanding, the ability, and the practice of

(1) knowing that there is a system mixer that allows access to sound levels of individual apps,
(2) knowing that it's possible to control the sound level of most individual apps in the OS mixer,
(3) actually does that in a way that turns down the normal OS sounds,
(4) in favor of turning up the overall OS level to support the one app (MS) that cranks out near-inaudible sounds at pp level,

is most definitely an advanced, _non-normal_ user.

For example, to me, you are _far_ from normal with your very deep understanding of sound levels in both music and computers (to say nothing of knowing that the ear is a non-linear mixer itself (which enables us to hear "beats" between frequencies)). I can't think of anyone in my computer or music circles who has anywhere near your understanding of these issues.

So I could be wrong, but it seems to me that the most practical default compression should be set so that everything from ppp to fff should be easily audible with the OS mixer and sound levels set for "normal" OS sounds, for normal users that watch YouTube audio videos, etc.

A good test (to me) is to just install MS on a variety of "normal" user machines, play those pp Moonlight Sonata scores for the users, and see if _they_ can hear pp easily, and/or if they think it should be a bit louder for comfortable listening. If not, I would adjust the default compression values to increase the MIDI levels.

At any rate, this is not my call, so all I can do is hope for something reasonable in the future that gives me a choice of compression maps to suit my machine, my music, and my ears... :-)

And I'm totally with you on your second posting where you address the problem of actually associating MIDI level numbers with the various dynamics. As long as MS has only one mechanism to associate those MIDI levels with dynamics, and as long as that mechanism is buried below user visibility levels (as it now), MS (you?) will have to set an association policy, have that mechanism implement the policy, and essentially force that policy on users.

The only way users could escape from the single MS policy is to do what you did to your computer (boost the OS levels for MS, and cut levels for all other current and future apps), or dig into the inspector and reset the levels for individual pieces.

It seems to me that it would be useful to distinguish between the ideas of

(1) What MS should try to be capable of (wide dynamic range), and

(2) What users should be able to do (I think they should be able to easily choose their dynamic range / compression mappings as named sets (eg. narrow 60-110, default 50-120, wide 40-126, super-wide 16-126), and

(3) What the default dynamic range setting should be (not what it currently is, I hope).

In many of our posts, I end up thinking that I'm talking about (2) and (3) (which you clearly understand), but that you use evidence from (1) to support (or explain) the policies and mechanisms used in (2) and (3).

Believe me, I see your points for (1), and agree with them. But I don't think arguments from (1) are good support for the more practical UI issues of (2) and (3). You move easily from 1->3 and back again in a heartbeat, but normal users start at (3), and only move to (2) with added depth, and then to (1) with extreme knowledge.

I think a happy medium for everyone would be to implement the mechanism of (2) (User-visible, and hopefully user-defined, named sets for dynamic mapping), and then pick a compressed default that will minimize user complaints about pp-fff sound levels. With a good section in the manual on the whole topic, people can then easily switch to a wider dynamic range if they want to.

And I appreciate you taking the time to discuss the issues in picking a policy. Let's make this thread a comprehensive one, now that we've invested so much time and thought.

PS. You say:

That is why we really cannot be in the business of assuming people will have
their system volume turned down low and therefore applying lots of
compression by default. It will pretty much destroy the effectiveness of
playback of dynamics

And I would respond,

By symmetry, neither can you be in the business of assuming people will have their system volume turned up high and therefore removing all compression and using a very wide dynamic range. Doing so would pretty much destroy the audible level of pp (which is where we are now).

PPS. I hope you are not assuming that my system volume is "turned down low", because it is not. I showed two screenshots, one with sysvol at 20% (or so), and one with it at 50% (or so). Whatever the level is, I just turn the knob on my external (high quality) physical speakers up or down until the sound level is comfortable. In both cases, I can't hear pp in the Moonlight.

The only way I can hear the default dynamic range is to do what you have done -- push the OS level up for MS, and adjust the sound levels for each app down, until things are comfortable.

Maybe an easier way for everyone is to just explain what users have to do on the first page of the manual -- turn up your OS level high, turn up the MS mixer level up high, and turn down all apps except for MS, until everything is comfortable. It's a nice solution in a way -- it's front and center and would reach anyone who looks at the handbook, it gets to the heart of a solution quickly, and it's really easy to implement -- just add the front page paragraph! (No messy code or policy or mechanism investments... :-)

A question worth finding the answers for is: are the two volume controls (slider on playback panel and the rotary volume control for each instrument in the mixer) already at 100% as a default? I can't make any assumptions as both have no a visible scale.

Earlier incarnations of the v1 play panel showed the scale going over 100%, and audio distortion will result when adjusting the play panel too high. I don't think the rotary volume controls on the mixer are at 100%.

Having these two set at 100% as a default would help the situation somewhat.

I agree, this is a good discussion to be having.

But to be clear - while you might say I am not "normal" in having such a deep understanding of how these things work and how to tweak them to produce what I want, I might counter that you are not "normal" in having spent money on external speakers so powerful that you need to turn down your master volume in your OS to avoid being blasted by the system sounds. I think more "normal" is to use the built speakers, and given how weak these tend to be, the "normal" thing is to have the system volume cranked to max most of the time. And in these settings - again, three totally different computers I used - "pp" already is very plainly audible.

Which is to say, if you just buy a computer, open it up and turn it on and start playing with it, the first thing you are likely to do when you go to play some audio is turn the system volume all the way up (assuming it didn't come that way out of the box). I have to believe this is going to apply to the majority of users. And when they then install MuseScore and play audio in that, it is going to sound absolutely correct already. "pp" is soft, "ff" is loud (as loud as their builtin speakers will go, anyhow), and there will be clear difference in volume between dynamic levels. Any significant compression will *worsen* the experience.

So it really comes down to trying to guess which is more common: for people to be playing through built in speakers turned up to the max, or to be playing through external speakers with the system and external volume controls set as you do.

Actually, though, it's really a simpler issue than that:

Does it really make sense for "pp" to mean "2/3 as loud as the maximum volume the program is capable of playing given the current system and speaker volume settings"? I say, no. The user has to take some responsibility for having his sound turned up high enough so that the default dynamic is a comfortable *medium* volume from which you can go up without going deaf or down without straining to hear. If a compressed option is provided for the benefit of people who prefer their system volume to be set so low that "pp" is inaudiable, I just don't think it should be the default. It defies common sense for "pp" to create a sound level more that half of what the system is physically capable of producing.

schepers - good question, and I don't know the answer

It's kind of funny that you think my setup is not normal. I run a routine desktop machine I bought off the internet, and $20 Logitech external speakers (because the original computer didn't have any speakers at all.) I use the onboard sound too, no special sound cards.

And I don't think I run my volume that low. Like most normal users that I imagine, I listen to the odd bit of music, talk to people on Skype, don't like to be shocked by loud system error beeps or incoming email dings, etc.

It just seems to me that MS doesn't play nice with other apps, or with the OS sound levels either.

Even you admit you have to crank up the OS and crank down the other apps to make MS sound "right" (whatever that is). I just think MS should play nice with the system and with other apps, and should sound reasonable while doing so. And if that means defying some inner sense of musical logic that sophisticated MS users have, so be it. Let the advanced users pick a wider dynamic range option on their systems. It's easy for them to do, and essentially an inaccessible thing for the average person to do.

I'll say it again. It's not my call.

All I can do is contribute my two bits as yet another "normal" user who couldn't hear pp out of MS, and who was essentially told that my computer, my apps, and my understanding of dynamics are at fault, and that the right and workable "solution" is to change the sound levels of my speakers, my computer, and all other apps on it in order to allow MS to "do the right thing" with dynamics.

In closing, I don't think we're making any progress here. I do not argue with your logic of pp = 2/3 of the max being illogical. I merely say that it is workable and audible for default users.

And whatever the outcome of this thread is, I think the policy decision should get a front page slot (a warning and a link to a deeper discussion) in the handbook.

I think you are misunderstanding me.

First, I didn't say your setup is "not normal". I specifically said "everyone has their own idea of normal". Your normal apparently has your system volume much quieter compared to mine. But my normal is just as valid as yours. Someone else might think both have things set too low; someone else might think we both have things set too high. There's really no way one group of volume settings can produce ideal results for everyone. If we optimized for people whose volume is set as low as yours (or lower), then people with their volume set like mine think dynamics just plain don't work. If we optimize for people whose volume is set as high as mine 9or higher), then people whose volume is set like yours will think "pp" scores are not playing at all. We have no way of knowing which "normal" is actually more common, but see below.

Second, I didn't say I turn up my speakers because of MuseScore. I turn my speakers up for music, period - any program that plays music. At anything much less than 100%, most music will sound very quiet through most computer speakers. Try it yourself and I'm sure you'll agree. That's why, I'll bet, you went out and spent $20 on better ones. I got in the habit of turning up my speakers years before I ever heard of MuseScore. Then I notice OS sounds were awfully loud, but I didn't want to have to turn down the overall system volume, so I simply learned how to turn down OS sounds.

So it's not that MuseScore doesn't play nice with other applications. On my system it does. MuseScore's "ff" is not unlikly any other program's "ff", and it's "pp", is not unlike any other program's "pp". I quite simply don't see the problem you do, because my system volume is higher. I can get away with a higher system volume because I turned down my OS sounds

Above I said we have no way of knowing if most people have their volume set low like yours (in which case they will think MuseScore is quiet), or high like mine (in which case they think it's perfect, but the change you suggest would make them think dynamics are broken). Do you see that we cannot satisfy both sets of users at once?

There are two ways of resolving conundrums like this. One would be if we knew what percentage of users had which type of sound settings - then we could optimize for the more common case, and make sure there is an override for the less common case. Well, we are already optimizing for my case, and there is already an override for yours: turn up the system volume. Providing an option to enable compression would certainly be possible as well. The other way one can resolve these issues is to consider which case has the more *obvious* solution. When faced with playback that is too quiet, I think most opeople have the sense to realize that turning up the volume is a solution. But if people started hearing dynmaics not work at all, they would have no clue why - they would just assume MuseScore stopped supporting playback of dynamics. So this also is an argument in favor of the current defaults.

Nio matter how you slice, I just can't see a reason to change the defaults. Common sense says "pp" needs to be a lot quieter than the maximum system volume - setting it to 2/.3 as loud as your computer and speaker are capable ofd producing makes no sense at all. The question of which volume settings is more *common* we can't really answer, but the question of which solution provides a more obvious workaround for the people for who the defaults don't work also very clearly points to keeping something like the current defaults.

Again, a *slight* compression wouldn't bother me. But if "pp" isn't very obviously a heck of a lot quieter than "ff", we have made things worse, not better. And I don't think there is a whole lot of room fto play with here - I thinbk you'd still perceive the vokume as low if your volume is really so low you can barely hear it now.

Consider - the OS sounds on your computer set the standard for "fff". "pp" needs to be six audibly distinct levels lower than that or pewople will say dynamics are broken.

I have a great deal of respect for your logical arguments. I agree with almost all of them. For example:

I agree with you that MS must map pp-fff to 0-127 MIDI levels.
I agree with you that some default policy mapping is required, else MS won't work.

I agree with you that no policy you chose will satisfy everyone.
I agree with you that defining "normal" is impossible (and unproductive IMHO).
I agree with you that everyone's normal is normal to them.

I agree with you that the current MS policy is logical - it does a reasonable, logical mapping of dynamics that preserves dynamic range of the scores that it plays, from ppp=16 to fff=126.

I agree with you that people will eventually "take some responsibility" and find some way to work their computer sound levels to get what they want (maybe by adjusting their mixer levels, OS sounds, apps, or speaker knobs).

I agree with you that the dynamic range put out by MS often exceeds the dynamic range of the sound systems that MS is played on (you've said this in various ways -- turning up laptop speakers, volumes set too low, all amounting to sound systems that can't really do MS justice.)


My point is that it could be a LOT easier for typical unsophisticated users to have a better experience from MS if pp wasn't so low, and that it could be a LOT easier for everyone to choose their preferred dynamic range if they could just choose a range from an MS menu, like so:

Edit, Preferences, Dynamic Range:

(A checkbox could select one for MS to use):
narrow (60-100)
default (whatever)
wider (40-110)
widest (16-126)

And then add some doc (a warning/link up front, a detailed section in the handbook).

Here I am arguing for a better MS _mechanism_ (checkboxes, doc) for users to control dynamic range. I am not arguing in any way for a particular policy for MS to adopt. I say give users a better mechanism (doc and some drange checkboxes), and let them choose their own policy.

One final post for me on this thread, after some experimentation.

I've been playing with sound levels and MS dynamics over the past few days, to see what I could learn. Here are my conclusions. The point of this post is try to provide some clarity of the key ideas at work in this thread. I hope this post adds positively to the discussion, rather than disturbing sleeping dogs.

1. My starting point is a "normal" machine like mine -- a routine desktop computer (no internal speakers, external Logitech $20 speakers), adjusted for "normal to quiet" volume levels for OS sounds (beeps, dings, etc), YouTube, ITunes, Skype, etc, such as might be found in an office environment where no loud music is allowed because it would offend your coworkers. A public v2.0 Beta of MS from a "front page announcement" in August was used (not a nightly build).

The main points here are that the machine (1) does not have its volume turned up loud for music lovers, and (2) especially does not have any OS mixer adjustments made to favor music apps over other apps such as OS sounds, skype rings and voices, YouTube videos, or online TV shows.

I know that using the word "normal" will trigger antibody reactions in some people (because they'll get lost in the definition of what "normal" means for individuals). But I need some kind of label for a machine that is not set up to favor the dynamic range of MS, and so I called it "normal." Feel free to assign any label that you like.


2. As things stand with the default MS dynamic range settings, the ONLY way to hear pp on a machine like mine is to (1) turn up the machine volume until it is "loud" enough to make pp audible, and (2) take special OS mixer action to turn down the sound levels for "all" other apps on the machine, otherwise said other apps (like OS beeps) would be so loud, sudden, and jarring that they would be uncomfortably loud.

The lowest dynamic that I could reasonably hear (to distinguish notes) was "p" (49) on my machine, about 30 ticks below the default mf=80.

The highest dynamic fff (99) did not seem "loud" at all, about 20 ticks above mf=80. (This build only allows 2 digits in the sound level, thus the 99. Later builds supposedly allow 3 digits and 126, which would allow 46 ticks above mf=80. So the extra 25 ticks might make fff "loud" on my machine, I don't know.

At any rate, now if I want to hear the full dynamic range of a piece of pp-to-fff music, I do what Marc (and I assume, everyone else) does, which is to crank up my volume and somehow deal with the way-too-loud sounds from the OS and other apps while I'm listening to the score.


3. This means that I once again must agree with Marc when he says (about other people who report audio problems with MS because they can't hear pp either) "Inevitably they have the volume set too low...etc."

But I think this statement could be rephrased to be less about the person, and more about the machine, like so "Inevitably, the existing (comfortable) machine volume settings above zero do not provide enough dynamic range for the wide default dynamic range of MS and any music that uses a wide dynamic range that reaches down into pp and ppp."


So the summary is that MS dynamic levels (ppp=16 to fff=126) are way too wide for "normal" (see the definition above) computer sound levels that have not been turned up "loud" for one reason or another, or have not been specially adjusted in the OS mixer to provide a larger dynamic range for the MS app.

At least on my machine, the "normal" (default?) dynamic range runs more like 50-126; of course mileage could vary on other machines.

I see three groups of MS users to consider:

1. musicians who need to play off the printed notation
2. people who run adjusted machines and sound levels
3. people who run "normal" unadjusted machines sound levels

MS user musicians in group 1 need all the ppp-fff range that is allowed, since it's on paper and notation is the main point of MuseScore. So there's nothing to change there.

MS users in group 2 (adjusted machines and sound levels)
- want the dynamic range from their computer,
- know how to adjust all the MS levels (MS levels, mixers, play panel, synthesizer)
- know how to adjust the OS mixer levels (specialized treatment for different apps),
- and will do it

MS users in group 3 most likely don't have the knowledge to be in group 2 (I was a newbie this way when I started this thread), and may not want to be in group 2 even when they understand and are capable of giving specialized treatment to the MS app through the OS mixer (where I am now).

MS cannot simultaneously help people in both Group 2 (adjusted machines and levels) and Group 3 (unadjusted machines and levels) right now, and so has been designed to support only Group 2 (adjusted machines).

MS cannot support both groups 2 and 3 simultaneously because it does not have any software mechanism for doing so.

In particular, MS currently has a hardcoded set of dynamic levels (ppp=16 to fff=126) that users (especially newbie users) cannot change in any practical general way to apply (dynamic range compression) to all scores. So the MS dynamic range _mechanism_ (which is to hardcode dynamic level MIDI numbers) currently enforces a single MS policy for Group 2 users that says "If you want to listen to scores with a full dynamic range, you have to adjust your machine in order for this app to work the way we mean it to work, on music that uses low dynamic levels like pp."

Please notice the difference between the ideas of mechanism and policy -- flexible software typically offers mechanisms that let users implement a variety of user-chosen policies.

My suggestion is to provide a more flexible, user-controlled _mechanism_ to implement a variety of dynamic level MIDI number _policies_.

For example, by clicking a checkbox, a user could chose the dynamic range that they want to use on their computer (eg 16-126 the current range, or 50-126 for my machine, or maybe 70-110 if users wanted to custom define the top and bottom levels.)

MS internally would map (scale) the default 0-126 MIDI levels into the target range chosen by the user (eg with a range 50-126, ppp could be 50, pp 60, p 70, etc, linearly scaling up to fff=126).

That way both Group 2 and 3 users would have very easy access to a mechanism that would give them exactly the policy that they wanted to use on their machine. Different policies could easily be used on different machines, if they had multiple machines such as laptops and desktops.

I see this thread is still active, which surprised me. The status is not my call, so I will leave it alone, and will understand if it is closed. I don't even think it's a bug report any more. More of a "limitation by (inadequate) design," I think. Thanks to all who participated in the discussion. I learned a lot about dynamic levels, MS, and OS mixers from posting my original blues in this thread. :-)

You say you want to avoid debate about "normal", but you still persist in calling your setup normal and everyone else's, apparently, abnormal (you use the term "adjusted", but that implies has been changed from sort of default you consider "normal"). I will refrain from making that same mistake in my response. The word "normal" will not appear in this response after this sentence.

I will now add just a few comments for perspective and clarification:

Perspective: once you set system volume to make your Windows chimes and so forth levels you can live with, you have now defined the maximum possible volume MuseScore can ever hope to ahiceve. That is, whatever volume a Windows alert bell sounds, that is the loudest that "ffff" in MuseScore can ever hope to ahiceve. Everything else will have to be scaled down to that. If you have turned down your system sounds to the point where a Windows alter bell is a lot softer than a real "ffff", then "pp" is going to be a lot softer than a real "pp". This is just common sense. They only question is how *much* softer than a real "pp". The answer for now is, MuseScore's "pp" is exactly as much softer than a real "pp" as MuseScore's "ffff" - which by definition cannot possibly be louder than your have set your overall system volume to be - is softer than a real "ffff".

If you are in doubt about what a "real ffff" is, hire a trumpet player to stand as far from you as you sit from your speakers, and ask him to play high C as loud as he can. If your Windows sounds are a lot quieter than than that, then naturally, "pp" is also going to be a lot quiter than what a real flute player playing middle C "pp" would produce.

Clarification: regarding max volume, it is always 127. It was a bug in the Inspector only that misleads one into thinking it is 99 in the Beta build, but "ff" is still really 112, "fff" 126. So in 2.0 Beta 1, "fff' really does *sound* as 126, it just displays incorrectly in Inspector.

Clarification: in any talk of providing a possible "compressed" mode, we shouldn't really talk about changing the MIDI velocity levels, but about the mapping from velocity to actual volume. That's because velocity 40 isn't just *quieter* than velocity 80 - it also likely uses a different sample, since real instruments change their sound characteristics at different dynamic levels. Even if you want to artifically compress the actual volume such that "pp" is not much quieter than "ff" (so that it is still audible even with system sounds set so "ffff" is not in fact loud at all), you probably still want to use the *sample* of an instrument being played quietly.

I hadn't changed the status because as I have said, I'm not opposed to providing a compressed option for those who prefer to keep their system sounds quiet and also prefer to use the overal system volume to make that happen rather than the systemn mixer that is provided for this purpose. I'm not even opposed to doing a *little* of this by default. So I've now changed this to a feature request to be considered for some future release.

BTW, an addendum on my "perspective" comment:

If you are not accustomed to dealing with professional audio / music environments, I should point out what I am describing in terms of "scaling" is absolutely not the slightest bit different from how it is with ordinary music recordings - CD's, etc. If you put on a CD that has a trumpet playing a high C at ffff followed by a flute playing middle C at pp, the latter is going to be a *ton* quieter than the former, just as it is in MuseScore. And if you played that CD on a computer whose overall system volume is set such that Windows chimes are not very loud, that means the trumpet playing high C at ffff is similarly going to be be not very loud (it's going to be virtually the exact same decibel level in fact), and the "pp" will be practically inaudible in comparison. In other words, MuseScore is behaving exactly like how real music as recorded on regular ordinary CD's, LP's, M3's, and what have you sounds. If your system volume is turned up such that "ffff" is not very loud, "pp" is going to be practically inaudible. That's a fact of life that I think would be well-known to anyone who listens to recorded music - and most users of MsueScore will fall into that category.

I generally agree with everything Marc has said. If you want to hear dynamic ranges wider than what your computer is set up for, you must turn up the volume on your computer. No argument there. I agree with all the stuff about real instruments, real dynamic ranges, etc. etc. too (really, it's all true, so who could rationally argue with it?)

But if you don't want to turn up your computer and mess with the settings, or if you don't care if your computer "accurately" reproduces real dynamic ranges?

All I'm suggesting is that MS give users a way of compressing the 0-127 MIDI dynamic range into something smaller that fits the user's computer settings, rather than forcing them to turn up their volume and mess with various level settings. Set whatever default MS policy you like, but it would be really nice if you gave me a way to compress the MS dynamic range levels to fit into my existing computer environment.

I was very frustrated with my recent MS volume experience, and with the long "discussion" in this thread (that essentially amounted to nothing more than "if you want to hear dynamic level p in MS, you have to change your OS mixer levels for all other OS and all other app sounds, because MS is right, and all the other sound levels and apps are wrong."

So I went out and bought Finale PrintMusic 2014. It arrived today. I installed it on my computer, which runs sound levels that let me hear all the other OS sounds, Skype, YouTube, ITunes, etc at reasonable levels for me.

I installed Finale, loaded the demo Scott Joplin piano rag, clicked play, and listened to it all the way through. Sure enough, the default sound level volume was just fine (presumably mf, since there were no dynamics at the score start). Then a section marked f came along under the playhead, and volume jumped up noticeably. Next, a section marked p came along, and volume fell noticeably, below the original level (assumed to be mf).

The p volume level was easily heard, not like MS and the Moonlight Sonata experience that motivated this thread.

I cite this effortless Finale experience, containing p, mf, and f dynamics, to reiterate my points in this thread:

- MS does not play nice with other apps

- MS demands custom changes to OS / app mixer levels to hear p dynamics

- MS has no easy way for users to compress the 0-126 levels into the existing physical dynamic range of their computers

- and (at least in this case) it seems impossible to change the minds of MS developers that the default behavior of MS should be changed in this case

Wishing you all the best, since I now have a notation program that does the notation AND plays nice with the other apps on my machine. (PS. I just deleted it, too. Really, who needs the hassle?)

@kkkwj Thank you for sharing your insights. Very valuable. I'm very aware of the volume issue, we just need the proper time to address it. Many things to do. Thanks again.

I wonder where the dynamic-number pairs that musescore is using have come from.

If the logarithmic nature of hearing is such that, on a constant scale, the lows will get lower faster than the highs get higher, then we definitely don't want more space between the lows than we have between the highs. And that's what we seem to have, based on the numbers someone posted above.

If we put mf at 80, and fff at 126, on a constant scale, that's 15.333 between each dynamic. That brings us down to 19 (rounded) for ppp.

If we want to adjust for this logarithmic nature of hearing, then ppp should be something higher than 19, not lower.

So it does seem like things have been skewed in the wrong direction, somehow.

Status (old) fixed active

well, I disagree ;-) it is way too loud now, also a lot of clipping at forte already, not to speak of ff or fff.

That's what I've been saying. Our *max* volume is not the problem (or if it is, only by a small amount). We could perhaps use a *small* boost, but also, some "compression" of the final results, so there is not such a large differential between ppp and fff. We sacrifice some realism, so ideally have a way of switching the compression off, but maybe starting off with some compression makes a better experience for newcomers.

BTW, the compression needs to be an audio stage, not MIDI. We can't increase the MIDI velocity values for the various dynamic makrings without messing up the velocity switches / layers in soundfonts - or existing scores that incorporate manual velocity values.

Title Default and max playback volume too low Default and max playback volume much too *loud*

For me, on my system (Toshiba Chromebook 2 running Ubuntu 14.04 under crouton scripts from with ChromeOS - admittedly not a common configuration), the new setting is *painfully* loud and distorts noticeably even at "mp" dynamic, and at "f" and above music is completely unrecognizable - just a loud harsh buzzing mess.

This is after running "mscore -F", so I know I have all defaults for my synthesizer settings. The volume slider shows a value of 10 below 0 (which I guess translates into -10 even though no negative sign appears in the display). With the extra gain factor in the code set to 20, volume is just *crazy* loud. Maybe this is system dependent? For me, I find setting that value t oanything above 2 or so distorts in louder passages; 20 is just insane.

I guess it's more likely the effect depends on instruments used than on system details. Attached is the beginning of the first score I tried - a simple quiet piece for three winds plus bass. With the volume at the default "mf", it's painful to listen to for all the distortion (even if I turn my speaker all the way down), and it still distorts badly even at "mp". Forget about trying "ff" (but just so you can see why I find this so shocking, maybe give it a try anyhow :-). we're not off by a little bit - we're off by an order of magnitude.

Attachment Size
distortion.mscz 12.56 KB