i recently decided to make a trance mix of singles that i have (all original CDs, not mp3s). i stumbled upon one single i purchased recently and when i played it over and over i notice strange HF artifacts present. at first i thought maybe i was imagining things. so i decided to analyse the file in soundforge. soundforge reports a sharp cutoff at 16.5 kHz. maybe the source is vinyl? i don't know. to make sure it wasn't some kind of intentional lowpass effect which i have seen before on other CDs (i.e. enigma), i ripped all the tracks from the single and noticed all the tracks had weird HF artifacts and all cutoff at 16.5 kHz. this CD was bought directly from a well-known dance music label. the thing is, is the music on the CD from a lossy audio format? too strange but i can't tell really, which is why i'm posting parts of it here and hope to get some feedback. plus, notice how the levels are quite low. no, it wasn't mixed this way for sure because all the tracks are quite compressed. maybe wavgain or a similar processor? this is not my doing, it's how it is on the CD.

definitely from a lossy source

thanks for having a listen , i'm quite relieved that i'm not the only one who noticed this. the website where i bought this CD from, hooj.com (now city16.com), also offer WMA downloads for purchase. i really hope they didn't just take the WMA files they offer for downloads and put them on a CD-R and pass it as original

I just gave it a look in Cool Edit's spectral view and you can see hf dropouts in the region from 14 to 18 kHz. If you like, I can upload a screenshot. At least that's a strong indicator you are dealing with a lossy source.

On the other hand, without having an original to compare it to, it sounds nice imo.

please do. there is another remix on the CD with a lot of high frequency content, all of it destroyed. it's quite obvious a lossy encoder was used.

fuck it.. i'll return the CD.

Here's the screenshot.

Being in the upload forum sure is practical

You can substract the channels. On positions with loud content in the center of stereo image, this will discover awful artifacts if the source was lossy. (The same happens if you don't plug in your headphones correctly. On lossless sources you'll notice 1st that you can hardly hear the singer's voice, on lossy you'll notice the artifacts 1st.)

hmm.... i'm trying to figure out the block lengths here. i see short blocks of ~256, but i can't seem to catch the long blocks. they're between 1900 and 2700 samples. (2048? if so you've got low bitrate AAC, most likely. i have no idea what the WMA block lengths are)

oh... and i think block lengths are switchable per SFB. don't know if that helps (any devs around here know how WMA works? i wouldn't touch it with a 10-foot pole, unless it was a jousting stick, but somebody's got to know something about this format)

mugfunky: i really can't tell what format was used. WMA is my first guess because the website that sells this CD also offer digital downloads in WMA format. the artifacts stand out too much for this to be original. the graph gecko posted is also a strong indicator it comes from some kind of lossy source.

tigre: i'll give your idea a try tonight just to make sure.

jan s.: hehe yeah i ran it already on all the files and got the same result: mp3 decoded source.

MP3-decompressed music on original CDs (and even vinyl) is quite common in the last decade. Many starting musicians send their demo-records to labels in MP3 format, and some "lazy" labels master the singles and sampler albums directly from those MP3's.

MP3-decompressed music can also be seen on compilation albums - when a big label license tracks from smaller labels, some of those small labels can be lazy enough to send their tracks as MP3 files (maybe because they do not know how to attach big WAV files to emails or how to share them in any other way)

But if you see a high-frequency cut-off at ~ 16 Khz - this does not always mean that the track was extracted to CD from an MP3 file. Maybe the reason was that the mastering engeneers had to remove the sound of the ... TV or video monitor from the recording (the sound of the flyback transformer). Here and here you can read the story about that problem ( ~ 16 Khz high frequency pitch in studio recordings )

The horizontal (flyback) frequency can be 15625, 15734 or 15750 Hz depending on the TV standard (PAL, SECAM or NTSC). If the original recording was pitched up or down, it's value in the recording can be a bit different. Sometimes mastering engeneers find this high freq noise too loud and try to remove it. And the resulting recording may look like it was decompressed from MP3, but if you look attentively, you'll see the difference from MP3.