I am looking for writing a program to achieve this. The video: https://tidalhifi.com/in/video/lossless-explained demonstrates what I am asking about. Thanks in advance.

  • 1
    flip the phase of one; extract the difference.
    – Tetsujin
    Dec 21, 2014 at 18:26
  • @Tetsujin can you please elaborate on this?
    – mightwork
    Dec 22, 2014 at 10:54
  • 1
    The simplest way to extract the difference between 2 audio files is to invert the phase of one of them, then add the 2 files together [assuming they are otherwise perfectly synchronised]; what remains is the difference between them.
    – Tetsujin
    Dec 22, 2014 at 11:05
  • Ohhh.... finally a lossless subscription... win!!!
    – AJ Henderson
    Dec 22, 2014 at 16:03

2 Answers 2


That depends on what you mean by audible. You can invert the phase of one and add it back to the other and it will play only what the difference between the two is, however you will hear artifacts that might not have been detectable to normal hearing in the original file since they were previously buried underneath other sounds.

It is also important to make sure you have things perfectly in sync when doing this. Any variation in position will result in issues as will any additional anti-aliasing that was done on the files during processing, so it may not be an exactly accurate comparison.


Audible differences is something different than electrically measurable differences. The techniques mentioned (by computing the difference between the two signals) allow to measure differences between two signals but don't give real details about the perceived difference.

The correct way to evaluate audible differences is to setup a panel of listener, and have them run a blind test on several samples in random order. Note that the test might be run differently depending on the scale of the differences. That is not an easy task.

Here's an example : the ITU BS 1116 Methods for the subjective assessment of small impairments in audio systems.

  • 1
    I agree that is the way to evaluate how noticeable the difference is, but it wasn't clear if the OP wanted that or, what I took it as, something that would let him actually hear what has changed. That's why I included in my answer that it isn't really what the "audible" difference is, but it is the closest you are going to be able to hear directly. Still +1 for good info on how to evaluate how noticeable a difference is to end users. I suppose neither answer really produces what the OP's link has, which is basically a dramatization of the differences.
    – AJ Henderson
    Dec 23, 2014 at 14:54
  • Fair. The OP's link is a marketing message. Few people are able to hear the differences (given they have a sufficient quality loudspeakers/headset system) between a 320k aac and a PCM (lossless) codec in average. You can use dedicated signals (like castanets for example) that will stress out the artefacts of the aac or mp3 codecs but these are not meaningful in general.
    – audionuma
    Dec 23, 2014 at 21:46
  • “Few people are able to hear the differences (given they have a sufficient quality loudspeakers/headset system) between a 320k aac and a PCM (lossless) codec …” — that depends on the source material. AAC is designed to work well for some kinds of audio and not for others. Acoustic strings, for example, get horribly mutilated in such a way that anyone can hear it. Music producers are choosing sounds that get through AAC encoding with the least mutilation, leading you to think that AAC is much better than it is. Jan 22, 2016 at 10:56

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.