Which acoustic spectogram shows higher quality of audio file?

Question

I have three acoustic spectograms of different copies of the same song:

1. MP3; 320 kbps; 9.20 MB
2. M4A (AAC); 256 kbps; 7.46 MB; Converted from 1. audio file via iTunes
3. M4A (AAC); 225 kbps; 6.87 MB

I want to determine which of these copies have the best audio quality. Since human hearing range is commonly considered to be within 20 to 20,000 Hz and that a 16 kHz cut is noticeable in the first two copies I would go with the third audio file harbor a superior audio quality. The 3. audio file kind of uses the human hearing range to the fullest while the first two ones cut off apruptly. This looks very artificial to me.

1. Is this correct? Are my assumptions correct?
2. Is there a software or a method which compares and finally determines which lossy version is of higher audio quality?
3. Can someone give me a profound explanation on how to determine a good 320 kbps MP3 file. (e.g. I get a MP3 which states 320 kbps and I think this is very good quality but do not know how this file got compressed at the first place, taking into consideration that LAME is a superior encoder than Fraunhofer)
4. Is it correct that M4A (AAC coding standard) is superior and generally achieves better sound quality than MP3 at the same bit rate?

Please correct me if I am wrong with assumptions. Merci

Answer 1

What you are asking is impossible to determine from a spectrogram. Spectrograms indicate frequencies present, not the quality of the audio. There may be loss of fine detail in one that would barely show up on a spectrogram at all or there could be noise and artifacts introduced that would make the spectrogram look more full.

Spectrograms do not differentiate irregular noise from signal. All we can say for certain is that the third one has much more information available above 18khz, which may very well make a significant difference for many listeners, but the rest of it could be horrible quality making the overall gain from the 18-20khz band a minor issue compared to other issues.

Answer 2

Best audio quality is difficult to assess objectively.

What you can say from the given audiograms is (as you mentioned) that N° 3 includes more high frequencies, which is usually considered better.

About the N° 2 audiogram : there's usually no benefit in transcoding a lossy file into another lossy file codec because whatever information has been lost in the initial lossy coding is lost.

Point 2 :

Currently, assessment of the quality of lossy codecs at a given bitrate is conducted thru statistical tests with actual people, as there's more than raw measurement figures involved.

Point 3 :

When you have a file encoded as 320 kbps mp3, you have no information on the source file from which it was encoded (not to mention which encoding library was used, although it might be mentioned in the file's metadata). The source file could be a 24 bits / 48 kHz pcm file as well as a 96 kbps mp3. I don't know of a process that would be able to determine what quality the source was or which encoding library has been used (unless this is part of the file's metadata).

Point 4 :

It seems that at bitrates over 128 kbps, the perceived difference between aac and mp3 gets negligible. See http://opus-codec.org/comparison/.

There are other points to take into account : mp3 is more widely supported than aac (think automotive for example). And the mp3 patents have now ended.

Answer 3

I think due to the highest data content among your 3 audio files of different format 3.AAC>1.MP3>2.AAC because the 3.AAC has highest data content and 1.MP3 has a few more data content than the 2.AAC, the 256 kbps m4a which you convert via itunes.