+ Reply to Thread
Results 1 to 4 of 4
Last edited by maudit; 21st Jun 2023 at 12:39.
LISTEN to them.
There are some things you can gather from spectrum readouts, like relative volume or lack, edits or lack, and brute filtering or lack, but "quality" is not one of them.
As an example, you can have an uncompressed master and a compressed copy, both highcut filtered, and they will look very similar (though not identical). But you can (or SHOULD) be able to HEAR the difference.
As Cornucopia pointed out, you can't really tell audio quality by looking at a spectrogram (except maybe with very extreme differences). The spectrogram on the left shows higher frequency extension -- 24 KHz vs 22 Khz. But that could be all distortion artifacts, not part of the original signal.
Generally, audio codecs roll off the high frequencies to get more compression. The lower the bitrate the more the high frequencies are attenuated. If you are over 30 you can't hear frequencies over 20 KHz so you won't be able to distinguish that difference in the above audio samples. As you age you lose those higher frequencies. The older you get the more you lose. Anyone over 50 or so should recuse themselves from judging audio quality.
My 2 cents sightly different than two previous statements - spectrum (spectrograms) can be useful tool for some audio distortions aspects so based on spectrograms quality can be evaluated and compared but i agree that this not miracle tool to judge audio quality in general. Spectrum (spectrogram) can't replace listening and can't be used as only single tool to evaluate audio (and signal) quality - this is one of few mutually complementary quality analysis tools.
btw - right spectrogram obviously show some problems with frequency linearity.
In theory you could perform convolution to get some informations about distortion nature but in real life this is not so easy task.
Last edited by pandy; 21st Jun 2023 at 14:37.