Universal MG embed an audible watermark in downloads (article@post#30), Was: FLAC file with artefacts not present on original CD? Options

[mannheim]

This is my first post in these forums.

Can anyone help me guess the source of some audible artefacts in some downloaded music?

I have two FLAC files for the same classical track (Tchaikovsky's 5th Symphony, Markevitch, on a Philips label). One was purchased from Passionato, the other was purchased from UMG's web site. Although both are supposed to be lossless copies of the data on the original CD, the FLAC file from Passionato actually has a clearly audible sort of "flutter" in some quiet, sustained passages. It is not the same as the UMG file. The difference isn't subtle. (Having read the TOS, I did do an ABX test. But it hardly seemed necessary in this case: I got 10/10 without difficulty. I will upload the short samples that I used.)

My guess was that the Passionato download was actually derived from a compressed audio file, converted to FLAC. But I'm puzzled. Compressing the UMG file and then converting back to FLAC did not give me a result that was in any way comparable, whatever compression I tried. The UMG and the Passionato file have exactly the same number of samples, and the difference of the two wave forms has a spectrum which is mostly in the range of about 1000 Hz - 2500 Hz, with very little above that.

Another possibility is that both FLAC files were obtained from some (fairly high quality) compressed source -- say MP3 -- but that the Passionato file was decompressed using software with a bug. I have quite a few FLAC files from Passionato. All the files from UMG labels exhibit this "flutter", with the exception of the most recent one I purchased.

I will edit this post when I have a link to the samples. I'll upload 7 seconds of each of the two FLAC files, and also the difference of the two wave forms (amplified). The difference sounds like fragments of the original tones together with higher "chirps". (I read the TOS concerning differences of wave forms, but I think the difference is interesting in this case.)

Edit: The uploaded FLAC samples are here.

This post has been edited by mannheim: Jul 21 2011, 18:56

[mmontag]

I have some analysis of the UMG watermark here in case you guys are interested, and so that there can be no more question about the origin and nature of these artifacts. The watermark technology they use is criminally bad.
http://www.mattmontag.com/music/universals-audible-watermark

[Porcus]

The watermark technology they use is criminally bad.

In terms of what? Is there any technology to compare this to, or is it so that this could very well be the state of the art?