I am about music audios.

These can be compressed by different formats with different degree. (I.e. I don't mean the dynamic compression).

After being compressed by a large degree (say 96kbps mp3) the audio may have been saved at a lower "fake" compression degree (say 224 kbps).

Is there a way to estimate how much compressed an audio's content really is (at the above example: telling from the 224 kpbs that the content is not more than about 100 kps)?

The characteristics I calculated so far (freq distr, phase shift etc.) show no difference between the different compression ratio's results.

Now I was thinking from the codec side: Mostly psychoacoustic probably. So... maybe searching for its typical effects: masking, impuls behavior etc.?

Are there any experience, hints, ideas what to calculate?

I can't see any connection to the suggested answer of that other thread. (It's about finding a correlation of a signal to a given pattern (snoring in that case). But my question is completly different: I need to know if there are any characteristics in a signal to tell that is has been compressed by codec and then "blown up" again. I don't have any "master"; neither of the wanted signal, nor of the noise/ snoring. Just an audio of which I want to know if it's as good as its technical specs or if it's worse, as it has been much more compressed before, and unknown by me.)

  • $\begingroup$ Ah, I misclicked. This might help you: dsp.stackexchange.com/questions/48027/… $\endgroup$ Commented Sep 13, 2021 at 14:46
  • $\begingroup$ And you'll definitely want to read both of these answers: dsp.stackexchange.com/questions/58816/… $\endgroup$ Commented Sep 13, 2021 at 14:46
  • $\begingroup$ Thank you. I'll have a CLOSE look at that (will take some time). Then will be back for reaction. $\endgroup$
    – User42
    Commented Sep 13, 2021 at 15:04
  • 1
    $\begingroup$ Meanwhile I had the close look I needed - but none of the threads helped me much. Saying not more than "It's hard to tell". Least of any help is the thread that my question has been marked as duplicate of. I don't see any connection to autocorrelation. Actually I'm much concerned how anybody could decide for me that MY question is answered. $\endgroup$
    – User42
    Commented Sep 15, 2021 at 11:17

1 Answer 1


If you know the reference input you might reverse engineer codec behaviour by trying all different codecs at all different bitrates until one matched. Or you might se telltale signs (rectangular patches in the frequency domain where energy is just above/bellw the threshold of being/not being transmitted) of a filterbank that is known to be used for one particular codec.

But in general, it is a hard problem. Codecs are allowed to do any linear le nonlinear operation that they think will make the audio sound «similar» but the bitrate is decreased. An audio-to-midi-to-audio chain would be a hypothetical example of (extreme) compression. How would you detect that given only a sample of its output?



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.