https://pdfs.semanticscholar.org/5997/b ... 876cad.pdf (https://pdfs.semanticscholar.org/5997/b358206f894f3a1571e1c4a8af68f9876cad.pdf)
Basically you can take a loseless audio file and subtract (phase invert) the compressed file (mp3) render that to a new file and your left with the frequencies that usually get lost during compression.
There are two methods he used to achieve this:
1.Take a wav and a mp3 (same file), phase invert the mp3 and render to a new file.
2. And "zero the corresponding bins in the original uncompressed file where the difference is null or near-null, i.e.- using the MP3 as a mask on the original file".
The First method is very easy to recreate with any daw, but I want to try the second one for different sonical results. But I have absolutely no idea how to achieve this and almost no coding experience at all (except visual programming).
He mentioned python and some libs for that, but sadly I have no details of that process
Anyone got an idea?