and sometimes vice versa. (Even taking into account the lighter blue shaded dynamic rather than the peaks)
(In any version of Audacity)
For example I recently recorded a multi track composition that appears in Audacity waveform display as quiet looking but shows up as a pretty high volume in the db display. Indeed it plays back with a perceived loudness equivalent to a rock track I composed which shows a far higher waveform - almost clipping.
What accounts for this?
In the amplication effect it suggests an increase of 10db to bring it up to zero but if I were to do that (or anywhere near) the volume would be excessive.
Sorry, I should have mentioned that I was referring to the final exported mixdown wav.
The obvious shift is between the waveform peaks and the RMS value. They don’t have to track. Any performance with leading percussion is going to have serious peak values compared to overall loudness. You have a very serious problem if you have RMS higher than peak.
Add the background noise value and you have the three AudioBook settings that ACX checks for when you submit your work for publication. No fair having the relationships all jacked out of shape. They’re required to be in narrow ranges in order to keep chapter by chapter and book by book consistent and enjoyable.
Since you’re on Windows, are you sure Windows isn’t trying to "help you?? I never tire of telling the story of how the company gave me a computer to check for audio quality. It wouldn’t check. It would not pass. Turns out it came out of the box with Microsoft Cathedral Sound Effects running and it wasn’t obvious what was happening.
Do you leave Skype running in the background? You probably shouldn’t. Skype loves to reset your computer sound settings to suit itself. That’s how it got the reputation of “always working.” Unfortunately, sometimes that causes other programs not to work. It doesn’t like music very much and Windows Enhanced Services doesn’t, either.
http://manual.audacityteam.org/o/man/faq_recording_troubleshooting.html#enhancements
Koz
Thanks koz, no I don’t have skype installed and windows “enhancements” are disabled. I have a guide I use for optimising a pc for audio recording and playback.
I’ve noticed this several times on different computers with major studio releases as well as my own stuff. With some tracks waveform height doesn’t necessarily conform to listening volume.
Here are screenshots of waveform and waveform db of the same track.
It doesn’t need to be any louder to confortably match the listening volume of anything else I have.
I don’t see anything unusual…
Perceived loudness is complicated. It’s related to the (relative short-term) average/RMS levels and the frequency content.
Loudness is poorly correlated to the peaks. Our ears don’t respond instantly and you’ve probably already noticed that all 0dB normalized files don’t sound equally loud.
Our ears are most sensitive to mid-frequencies (around 2kHz) and the ear’s “frequency response” changes with loudness (you can look-up “equal-loudness curves”).
Indeed it plays back with a perceived loudness equivalent to a rock track I composed which shows a far higher waveform - almost clipping.
Modern commercial recordings are compressed to death ([u]Loudness War[/u]). Dynamic compression brings-up the overall-average level without boosting and clipping/distorting the peaks. Or, sometimes they do push the levels into clipping.
Homemade recordings are rarely as loud as commercial recordings. Even if we use compression, most of us are not expert enough to squeeze-out the kind of loudness a pro mastering engineer can get. And, we may not want to compress the life out of our recordings! 
If it sounds right, it is right.