Low Volume Exports

• Jun 26, 2016 - 17:05

Hi fellow composers,

I'm having an issue with MuseScore which one of you might have the wisdom to resolve. WAV files I export from MuseScore, even with dynamics maxed on all instruments, are deficient in volume by at least 5db compared to standard audio files (e.g. songs I have downloaded). I have tried boosting the volume of some files using the amplify tool in Audacity and the audio gain tool in Adobe Premiere, but this results in a loss of quality. Is there any way to resolve this?

Comments

• Jun 29, 2016 - 17:16

The waveform on the bottom has been very heavily compressed - which is to say, the soft passages have been artificially increased in volume. The loudest passagesare the same, but the compressed basically has no softer passages at all. This is pretty common for pop music I guess. You are free to run your audio file form MuseScore through a compressor as well - there is one built in to Audacity.

Reply

• Jun 29, 2016 - 22:54

I could, but any audio editing tends to result in crackling on the file; I downloaded Polyphone (I think that is the name) to edit the soundfont and give it volume gain - but thus far attempts have been unsuccessful. As I said dynamics are maxed out and volume levels are deficient. I listen to songs on volume level 10 on my phone; for music I export from Musescore I up the volume to about 13, and 15 when outdoors. Any other way you know of to correct this issue?

Reply

• Jun 29, 2016 - 23:24

Adding compression should not result in "crackling" - like I said, it's added to most pop music already. Perhaps your first attempts just used bad settings or a not good program. Audacity does the job well. There thing is, even though you've set dynamics to the max, there still needs to be headroom because if you add more instruments it gets louder. So far fort ine for can't be as loud as ffr for 137 electric guitars. That's why virtually all professional recording adjusts levels in post production.

Reply

• Jun 30, 2016 - 08:33

Audio exported from MuseScore should already be compressed. The export happens in two passes, the first one finds the peak of the audio, the second one export the audio and apply a gain of 0.99 / peak. See https://github.com/musescore/MuseScore/blob/60bcbcb931319f83001a1cb6458…

So you can't really compress it more. The rest of the music we are used to listen has probably a smaller dynamic range.

Reply

• Jun 30, 2016 - 17:33

I could be wrong, but it seems what MuseScore is doing during output is more "normalization" than "compression". That is, the dynamic range is not reduced; everything is just increased in volume so that the peak hits the maximum. That still leaves a very large difference between ppp and fff, unlike most commercially recorded music, and a more significant difference between fff played on one instrument compared to fff played by seventeen instruments, etc. Even things like a single note decay - like how a piano note gets quieter rapidly after you stri,e the key - is affected by much commercial compression, leading to a perception the piano is louder overall if compression is applied even when playing a series of single notes.

Reply

• Jun 30, 2016 - 09:19

Commercial music you get from iTunes etc has gone through a process known as mastering.

Since the advent of the CD this has partly involved making the track as loud as possible. Often actually damaging the waveform in the process, although this is not the case in the example you present.

The main tool used for this is a multiband compressor. This enables the mastering engineer to isolate a frequency band in the waveform, apply compression to it, and raise or lower it's level if thought necessary. Professional multiband compressors usually have 5 configurable frequency bands.

Once this process has been done the track is then normalised to a specific db level in the case of your example, it looks like 0db or near to it.

In general modern mastering engineers master far too loudly in an effort to get their tracks noticed through the brickwall limiters used by radio stations. In the days of vinyl they were restricted as too high an overall db level would make the needle jump out of the groove.

Sadly Audacity does not come with a multiband compressor as standard, but there are several VST multiband compressors around that you could install in Audacity's plugin folder. Most of the free ones are only 3 band, but Reaper's free audio suite contains one with 4 bands as standard and the ability to add more bands if required. There is also 5orcery which has a unique take on multiband compression.

Hope this helps you to achieve the sound you are aiming for.

Reply

• Jul 2, 2016 - 15:20

Hi,

Haven't applied the compressor yet; read an article which asserts that it is best to apply mastering to 24-bit files, but the .WAV files that Musescore exports are 16-bit. I can export them from audacity as 32-bit files, but not as 24-bit files. Any way to remedy this so I can start mastering? Or, in your experience, is it necessary at all to use 24-bit files for mastering in Reaper?

Reply

• Jul 2, 2016 - 21:22

Downloaded Adobe Audition and have been utilising the mulit-band compressor which it offers. Results are impressive thus far. Audio is now nearly at the levels exhibited by songs I have downloaded which themselves have experienced considerable compression.

Reply

• Jul 3, 2016 - 09:26

There is a lot of nonsense talked about 24bit.

In the case of mastering it only makes sense to master in 24bit if the recorded sources are 24 bit.

In the case of MuseScore the soundfont samples are all 16 bit so to convert the audio to 24 bit for mastering is absolutely futile.

So why is it said that 24 bit is better?

It is because of the additional headroom recording in 24bit gives you before digital jitter affects the signal. This is particularly relevant when recording low volume sources such as quiet passages in classical music, as the noise floor is lower in 24 bit than 16.

However, for most purposes 16 bit is good enough - let's face it, we though 8 bit was good enough in the 80's for sampling :) But nowadays those samples sound really crunchy!

If you want to know more about it and why the noise floor is lower in 24 bit I suggest you start with the following Wikipedia article........
https://en.wikipedia.org/wiki/Digital_recording

I hope you haven't paid money for Adobe Audition and that you are using the trial edition!

IMO your money would be better spent on DAW software such as Cubase or Sonar where you will get more for your $200 or so. That will then give you the opportunity to render your MUseScore creations in MIDI, load them into the DAW and apply further editing, effects, etc.

There is very little more that can be done in Audition than in Audacity, particularly if you add VST plugins.

HTH :)

Reply

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