Playback is VERY Quiet

• Jun 11, 2019 - 03:58
Reported version
3.1
Type
Functional
Frequency
Many
Severity
S4 - Minor
Status
closed
Regression
No
Workaround
No
Project

Hello! I've been using MuseScore 3 for a while now, and I'm noticing that its playback is MUCH quieter than MuseScore 2. This is strongly confirmed by exporting the MP3 file and running it at the same volume. I have attached an MP4 with audio to show this difference. However, when I set the mixer to 0.00db OR the synthesizer higher, I can get it to the same volume level, but it becomes extremely distorted (where on the MP3 file it is not). I have no idea how to provide reproduce steps for this. Setting the volume on the instrument from 100 to 127 does make a difference, but not a big enough one.

Attachment Size
2019-06-10_20-53-43.zip 3.42 MB

Comments

Hmm, this could depend on your system, or if you are using a custom soundfont, or maybe the specific instruments you are using. But in general, there should be a noticeable difference at all. I just did a comparison using one of the OpenScore scores (Bach Brandenburg Concerto #3 ) and there was no discernible difference.

Status active needs info

I'm confused by what I'm seeing in the video - I don't see you running MuseScore 2 at all? - but know you can't compare the playback within MuseScore to an mp3, because audio level compression is applied. So if you're saing the live playback from Musescore is quieter than the mp3 export from MsueScore 2, that's true - and so is the converse. It's not a 2 vs 3 difference but a live playback vs mp3 export difference.

In reply to by Marc Sabatella

I think you appear to be missing the point. I'm not caring much about the audio quality, nor where it is coming from. I did not show anything important in the video, you must listen to the video. I guess if I were to properly name this issue, it'd be "Distortion at high levels of volume," because either way, MuseScore is quieter than ANYTHING else I play on the PC regardless of the format (mp3, mp4, ogg, .mid using an external player, YouTube, Pandora, Spotify, Discord, MuseScore is QUIETER). This is clearly an issue with MuseScore not matching/not being able to raise the volume to proper levels; I would raise my overall volume, but then the synthesizer gets extremely distorted (something that does NOT happen with my external synthesizer, OmniMidi; and no, I won't just output the MuseScore audio to that, this is an issue with MuseScore itself).

I did listen, but unless I am missing something, you never played the score in MuseScore 2. It appears you compared the live playback from MsueScore 2 the MP# export from MuseScore 2. That's apples to oranges.

The reason MuseScore - both 2 & 3 - sounds quieter than other things on your system is specifically because the compression I mentioend that is applied to most MP3 and WAV files. You're accustomed to hearing things normalized, but live performance doesn't work that way. The score you are playing can't be as loud as other things, because then scores that are inherently louder - say, a score for 100 instruments all playing fff - would not be able to be any louder than the mf solo piano score you just played. Again, MP3 files are normaized so indeed, mf solo piano is as loud as fff symphony orchestra, but it won't be that way in live performance from within MuseScore.

And indeed, cranking the voume up is not going to be a good idea, you'll get clipping within the synthesizer.

In reply to by Marc Sabatella

I understand what you're talking about, but while I did use MP3 as a comparison, it should not make a difference. You are STILL completely missing my point. I shouldn't have even mentioned MuseScore 2 because clearly it's confusing you. I have many issues with your previous comment.

First of all, yes MP3s are normalized, but not to the extent you're making out. A mf solo piano is not even close to as loud as an fff symphony orchestra. Even if this is an extreme example, the volume changes within an MP3 just as similarly as it does using a live editor. What is your excuse here? The fact a "live performance can't be as loud?" I beg to differ. Have you ever used FL Studio? It's professional editing software that allows you to create music of all types. It also supports MIDI. I have never had any volume issues with FL (in fact, it's actually louder than I prefer, so I always turn it down). Now don't try and make the comparison "but this is open-source software, not professional", and I get that, but my point of making this comparison is that LIVE EDITING SOFTWARE has NO EXCUSE to be quieter than the output volume of an MP3.

To add on to that, MIDI is also a live performance using a synthesizer. Using OmniMIDI, a MIDI synthesizer, it is marginally louder than the output of MuseScore. And this is, under your words, a LIVE performance as the notes are being mapped and then sent to a synthesizer (the SAME THING that MuseScore does). The MIDI volume is directly mapped to volumes like "mf" and "ff," so don't try and make the point that it's also being modified because it's MIDI.

If you don't understand my point, I apologize, but please stop trying to give these counterpoints that have no backing. Your counter points are easily debunked with multiple other programs, performances, and synthesizers. The POINT of my issue is NOT the differences between MP3 & Live, it is NOT the difference between "ff" and "fff", it is NOT the difference between MuseScore 2 and MuseScore 3. My point of this entire issue is that MuseScore IS QUIETER than EVERY OTHER PROGRAM. This is a simple issue of the output level of MuseScore, not any specifics of what's going on. The synthesizer should be much louder without distortion, which is proven that it can happen as shown using OmniMIDI (a MIDI SYNTHESIZER). MuseScore's built in synthesizer is quieter than even WINDOWS DEFAULT ONE!

Apologies if this was aggressive, but I'd just like to help identify an issue that's happening instead of trying to avoid the fact there's an issue in the first place, because there clearly is.

EDIT: Once again, I'd really like to apologize for being aggressive. I'm a very debative person and I can often get a bit out of hand when I disagree with something >.<

If I were to rename this issue, it'd be "Synthesizer distorts at high levels of volume," as I think that is the culprit of what's going on here. When I switch from a "p" or "mp" to a "mf" or "f", it quickly becomes distorted and is unable to handle the higher volume, resulting in the only solution to be turning it down, therefore making it quieter than the other applications.

OK, now that I understand you made a mistake mentioning MuseScore 2, we can forget a good chunk of what we both wrote. Also, apparently, you realize that mentioning MP3 was a mistake too, so we can forget wverything written there too. So let's start over.

Please describe, starting from scratch with no reference whatsoever to MuseScore 2 or MP3, what you perceive the problem to be. Feel free to reference FL Studio, but do so use specific examples as you did previously. That is, let's compare the actual decibel levels of MuseScore's playback with FL Studio's, using the two specific examples at hand (solo piano playing mf versus symphony orchestra playing fff). Ideally we'd then also go to a concert hall during a performance that features both and measure the decibel levels there as wel.

My best guess is what we will find is that both playing the fff symphony at roughly the same volume, but FL Studio will play the mf piano example louder than MuseScore does. Then, we can talk about which is more accurate (that is, which more faithfully captures the actual difference that would be heard in live performance), and then we can also debate whether the goal should in fact be accurately playing dynamics, or perhaps something else).

But that's the kind of data we'd need to go further. There can be no "debunking" of anything until we understand exactly what is being talked about here. Assuming my guess is correct and the experiment proves that MuseScore is more accurate and that FL Studio is applying compression to artificially reduce the volume differences, we could then simply agree to disagree about whether that is a good thing or not. That is, I expect you'd say yes it's good, and also you like Pepsi, I'd say no, it's bad, and I prefer Coke, and there would be no right or wrong about any of it, just a completely valid difference of personal preference.

BTW, digital clipping is indeed a thing, I don't think there is an issue here, but feel free to start a separate thread about why you think, based on your analysis of the waveforms, why it shouldn't be happening in some specific case where it is, and then we can proceed from there.

I just realized that the issue was due to Windows DirectSound vs Windows WASAPI. DirectSound has much more distortion and does not handle higher volumes well, where WASAPI handled it perfectly (hence why I wasn't able to reproduce the issue).