Detecting the real quality

I have this titanic task to do: I would like to run evaluation through all of my files and see, what is the real quality of them

I know there is BITRATE tag field in Mp3tag available under information fields sub-menu, but this gives only the supposed quality. [There also is the CODEC tag field, but that gives only info about codec used for creation of the file]. And it is useless for me because all of my MP3's at some point had same makeover done on them [after which they were encoded into 320 MP3]

But recently I got acquainted with the function of spectrogram in Audacity: by using it I can see if for example a FLAC file really have all those sounds not present MP3 files or if someone only encoded sound of lesser quality into a higher quality format. Unfortunately, right now I do this by hand, one file at a time; and it takes a lot of clicking every single time; and on top of that I also don't know how to export info about the real quality into a tag field

And as the difference between maxed out 320 MP3 and a FLAC is undetectable for me, a difference between 192-256 and 320 MP3 is. So If would want to replace the lesser ones with something better [320 or FLAC]; but in order to do that effectively / systematically, I would first need to single them out [and create some kind of a work schedule]

So the question is: what can I use [to do this in a kind of an automatic way]?

I've already googled this topic to some extent, coming out with the aforementioned Audicity software [and some other pieces of software which didn't stay too long in my system]. Maybe someone here already did the same thing with his / hers files? Or know what could I try out?

Perhaps you can run a test with WinMp3Packer and convert CBR files to VBR with a low VBR-setting in WinMP3Packer.
The lower the effecting VBR is, the lower was the quality, I would guess.

It seems, that you are on the way to find out the golden formula for the term "quality", which delivers a universally compatible concrete value.

Try out the SoX and SoXI command line utilities at ... http://sox.sourceforge.net
Beside many other audio related functions SoX can create a spectrogram as a png file.
For the first view there are PDF help files from the SoX package ...
sox_14_4_2.pdf.rar (183 KB)
DD.20151117.1047.CET

sox_14_4_2.pdf.rar (183 KB)

Creating info in a graphical form wold not do me any good. We are talking bout 15 00 files. I would need an info in form of a tag, that I could easily see in a column in Mptag; I would mark them as "very good", "good enough", "not acceptable", with a simple one sign code [like "+", "0", "-"]

So for now I use my ears

It has come to my attention in recent months, that some types of music / sounds are more prone to audible damage done by file compression. Like for example an opera signing still can be enjoyable; or some bass guitar and keyboards layed upon them. But a full symphonic orchestra sounds like heard from behind a wall. Or high hat sounds just distorted bad. So for now when I hear something low quality I send it to the Audacity

And in something like 9 out 10 cases I am correct in my estimates; which means if I hear a bad quality then in the spectrogram I can see a proof of that. But that takes way too much time and totally disrupts my workflow

It seems that sometimes the spectrogram shows something completely different from what I'm hearing with my own ears: /t/17591/1