I’m not sure what your question is, but…
I have got many music files on my computer which have got this negative levelling by default, I haven’t tweaked with them in any way. For some reason many albums released in 2007-2008 have the tendency to have the volume level somewhere between -9.0 db to even up to -12.0 db.
The acoustic reference of zero dB SPL (Sound Pressure Level) reference is 0dB, which is approximately the quietest sound that humans can hear. So acoustic dB levels are positive and a rock band might hit 100dB SPL or you might hit 100dB in your headphones, etc.
The digital reference of zero dBFS (Decibels Full Scale) is the “digital maximum”* so digital dB levels are normally negative. You can push the volume louder but you’ll get clipping (distortion).
Since we all have a volume control and “my speakers are bigger than yours”, there is no fixed relationship between digital and acoustic levels. But, if you reduce the digital level by 3dB (a bigger negative number) your acoustic levels will also be reduced by 3dB (a smaller positive number), assuming you don’t touch the volume control.
Music (or regular audio material) “loudness” is complicated… The maximum digital level is determined by the peaks, but the peaks don’t correlate well with perceived loudness. There are lots of quiet-sounding 0dB normalized (AKA “maximized”) files. If you normalize all of your music all of your files will have equal peaks but some songs will still be louder than others…
Perceived loudness is related to the (short term) average and the frequency content. Our ears are most sensitive at about 2kHz, so 0dB a mid-frequency 2kHz tone will sound louder than a 0dB low-frequency or high-frequency tone. And, short-term peaks won’t be perceived as loudly as continuous tones.
There are algorithms for measuring loudness. SPL meters use A-weighting. Digitally, RMS (a mathematical calculation similar to averaging) is better than using the peaks. ReplayGain and MP3gain use a more advanced algorithm, and there is an international standard called LUFS.
If you apply ReplayGain or MP3gain, your songs should sound approximately equally-loud. But, since many songs (including quiet-sounding songs) are already normalized (maximized). That means in order to match volumes the louder songs have to be reduced. That happens with ReplayGain (and related) and some users are disappointed when most of their songs are quieter than before. But, if you have enough analog gain you can simply turn-up the analog volume control and everything should be good.
If your music is not loud enough after maximizing you can use limiting and/or compression (with make-up gain) to bring-up the overall-average loudness without boosting/clipping the peaks. That’s the secret to “winning” the [u]loudness war[/u]. But, compression (obviously) reduces the dynamic contrast and music can become very boring. And if you over-do it, it can sound like distortion.
- Digital-to-analog converters, analog-to-digital converters, regular WAV files, etc., are all hard-limited to 0dBFS. It’s as high as you can “count” with a given number of bits. Audacity can go over 0dB “internally” and some formats can go over 0dB, but since you can clip your DAC you shouldn’t make files that go over even if you can.