Audible watermarking (intentional degradation) in music downloads and streams

[rob martino]

I had been noticing something in a few albums I was listening to in iTunes and Spotify lately (particularly classical music recordings), an irritating "tremolo" effect as if someone had a fan running in front of the speakers (or a strong tape flutter). I initially thought it might be due to the compression Spotify uses but noticed the same thing in the identical recording on iTunes (which should be higher quality).

It turns out the audio effect is a form of "watermarking" intentionally put there by music labels owned by Universal Music Group to better track where their music is going. I suppose streaming audio subscriptions might not be expected to offer super audiophile quality, but to have an artist's recording intentionally contaminated like this is pretty lame- and it's REALLY noticeable. This artifact will also be there when someone pays for a full price album on iTunes, as well as high resolution "lossless" albums from legitimate websites. Only CDs and the downloads offered directly from the label's websites don't have the watermark.

http://www.mattmontag.com/music/univ...ible-watermark

I'm surprised there doesn't seem to be more of an uproar over this- the watermarking deteriorates the audio far more noticeably than any conversion from an audio master to an high bitrate MP3. Perhaps in many cases people just don't notice?

Here's a listening test in which you can determine how well you can hear it:
http://mattmontag.com/audio-listening-test/

[Scott Bails]

It turns out the audio effect is a form of "watermarking" intentionally put there by music labels owned by Universal Music Group to better track where their music is going. Only CDs and the downloads offered directly from the label's websites don't have the watermark.

http://www.mattmontag.com/music/univ...ible-watermark

This is a disgrace, IMO.

[spacefreak]

Seems that in a few years, only pirates will have the real quality audio!

[rcarlberg]

The artifacts appear on UMG tracks at Rdio, Spotify, iTunes, Amazon, and others.

Easy answer, don't buy downloads.

Stick to CDs.

[ronmac]

Easy answer, don't buy downloads.

Stick to CDs.

+1

[Dave (in MA)]

The shittiness imparted by the watermarking reminds me of how the stereo audio tracks sound on the 70-75 Genesis Box DVDs, though maybe a little more subtle.

[Plasmatopia]

That won't protect you from the evils of brick wall limiting.

[Rael]

Aren't ITunes downloads generally lossy? For what they charge, I think you're better off getting the CD. Sometimes, on places like Bandcamp, you can get both a CD and a download of the same album for something like 10 bucks. I use ITunes to transfer music to my Ipod but I can't imagine ever buying music from their store. But that's just me.

[JKL2000]

The shittiness imparted by the watermarking reminds me of how the stereo audio tracks sound on the 70-75 Genesis Box DVDs, though maybe a little more subtle.

Were those watermarked in some way?

[rob martino]

Yes, AAC is lossy but I personally can't distinguish between iTunes' AAC format and the original CD, even on good studio monitors. For that matter, I have a 256K MP3 and SACD version of the same album (a fantastic audiophile quality recording of The Messiah on Linn Records) which I cannot tell the difference between- in my opinion media format isn't the biggest contributor to overall sound quality by a long shot (I'd say at most it's 98% recording/engineering/mastering, 2% media format).

And while bad brickwall mastering is mostly a poor aesthetic approach, this watermarking strikes me as particularly deceptive, especially when it's being applied to legitimate "lossless" downloads and the customer expects they are getting a high resolution (better than CD specs) reproduction of the original master.

[rcarlberg]

Aren't ITunes downloads generally lossy?

Unless you're concerned about file sizes, the "lossiness" isn't audible.

[Dave (in MA)]

Were those watermarked in some way?

No idea, all I know is that they sound like crap. Any time I've brought it up in various forums, people seemed to always respond that they only listen to the CDs in stereo or the 5.1 tracks of the DVDs.

[ronmac]

Unless you're concerned about file sizes, the "lossiness" isn't audible.

It certainly is. The louder it gets; the flatter the sound.

[rcarlberg]

It certainly is. The louder it gets; the flatter the sound.

Explain please. Are you talking about some kind of audio limiter???

Or are you saying the frequency response improves at higher SPL?

[ronmac]

Explain please. Are you talking about some kind of audio limiter???

Or are you saying the frequency response improves at higher SPL?

When the music is compressed, it becomes more noticeable when it's played loud. The more compression, the narrower the dynamic range and the flatter it sounds.

[rob martino]

When the music is compressed, it becomes more noticeable when it's played loud. The more compression, the narrower the dynamic range and the flatter it sounds.

It's easy to get confused about terminology like this, but data compression as it relates to media encoding (i.e. converting a CD to AAC or MP3) is completely different than dynamics compression used in music production (which reduces dynamic range as you stated). These "lossy" data compression methods reduce data/file size by attempting to minimize redundant digital information and parts of the frequency spectrum that are determined to be imperceptible based on psychoacoustic principles.

[ronmac]

It's easy to get confused about terminology like this, but data compression as it relates to media encoding (i.e. converting a CD to AAC or MP3) is completely different than dynamics compression used in music production (which reduces dynamic range as you stated). These "lossy" data compression methods reduce data/file size by attempting to minimize redundant digital information and parts of the frequency spectrum that are determined to be imperceptible based on psychoacoustic principles.

Okay, thanks. I considered not using dynamic range, and should have gone with my intuition.

But, to my original point (besides the mis-labeling), there is still a clear difference with lower bit rates. And, it is certainly more noticeable at higher volume.

[Azol]

The more people will learn about this "practice" the better. Spread the word. Or soon you will face audio watermarking during live performances right in the club or concert hall:

Audio Watermarking for Live Performance
http://www.research.ibm.com/trl/proj..._EI03Paper.pdf

[GuitarGeek]

Okay, thanks. I considered not using dynamic range, and should have gone with my intuition.

But, to my original point (besides the mis-labeling), there is still a clear difference with lower bit rates. And, it is certainly more noticeable at higher volume.

Certainly with mp3's, I do notice a sound quality with lower bitrates, say 90 or whatever. I remember back in the early Napster era, I was downloading out of print albums and bootlegs from there. One that I downloaded, I think, was one of the Ozric Tentalces albums, one of the original self released cassette only releases, which I believe was mostly unavailable at the time (this being before the second set of CD reissues).

Anyhow, there were two different versions, I can't remember what the bitrates were (either 60 and 90, or it could have even been 90 and 128), but I did notice the one with the higher bitrate had noticeably better sound quality. I suppose it was like the difference between recording an album onto a cheap low budget, normal bias cassette, and using a good high bias tape like Maxell XL's or whatever.

I remember in the early days of lossless file trading, one website had a Grateful Dead show that was encoded as an mp3, and I forget what the bitrate, but it had that sort of modulation effect (presumably accidentally) in it. Jerry sounded like he was playing his guitar through a phase shifter or a Leslie speaker cabinet, only it wasn't just his guitar, and it was the entire band, like maybe Owsley had the entire mix running through a Leslie or something.

Spreaker automatically encodes Journey Of The Sorcerer's Apprentice as a 128 bitrate mp3. To my ears, it doesn't sound bad at all, again to me I would consider it comparable to listening to a cassette tape air check, like I used to do back when listening to the radio back in the 80's and 90's.

[Plasmatopia]

I can't see the point in watermarking a live concert, aside from some sadistic need to inflict audio unpleasant-ness on your audience. If you made an audience recording there would probably be other ways of determining where the recording was made (set list, stage banter, etc.).

I'd rather see R&D money put into some way of stopping people from photographing and filming whole concerts by raising their phones into the air and blocking people's view. Maybe some sort of laser that sweeps the audience cutting off the hands of anyone that reaches above a certain height. Since we're being sadistic.

[rcarlberg]

Certainly with mp3's, I do notice a sound quality with lower bitrates, say 90 or whatever. I remember in the early days of lossless file trading, one website had a Grateful Dead show that was encoded as an mp3, and I forget what the bitrate, but it had that sort of modulation effect (presumably accidentally) in it. Jerry sounded like he was playing his guitar through a phase shifter or a Leslie speaker cabinet, only it wasn't just his guitar, and it was the entire band, like maybe Owsley had the entire mix running through a Leslie or something.

Those aren't encoding artifacts.

[GuitarGeek]

Those aren't encoding artifacts.

Then what are they?

[rcarlberg]

Most likely tape machine misalignment.

[GuitarGeek]

Most likely tape machine misalignment.

Perhaps, except there was subsequently another version of that same show, distributed to the public as Shorten files, that didn't have have that phasing effect. Of course, that second version could have a different lineage, that didn't include the misaligned tape machine (or perhaps the tape machine had been re-calibrated by the time the second version was done).

I had the understand that was the whole point why everyone went to lossless file technology, because there was something about the encoding of mp3's (at least 2000) that caused that sort of "swishing" effect. I always had the impression that's what i was hearing on that one particular Grateful Dead mp3.

If that's not what the deal was, then what exactly is the hang up people have with mp3's? I typically don't have a problem with them, but then I typically use higher bitrates (usually 320 or 250). To me, the quality is near enough to the original as to make no difference (as Jeremy Clarkson might phrase it).

[ronmac]

If that's not what the deal was, then what exactly is the hang up people have with mp3's? I typically don't have a problem with them, but then I typically use higher bitrates (usually 320 or 250). To me, the quality is near enough to the original as to make no difference (as Jeremy Clarkson might phrase it).

At 320, most people won't notice the difference between the MP3 and lossless. But, from what I understand, most people are downloading lower bit rates. Another factor is that the MP3 players and headphones are more forgiving to the lossy than higher-end equipment.