This is my personal blog. The views expressed on this page are mine alone and not those of my employer.
Update, April 8, 2013: Can you hear it? Take a watermark listening test.
A while ago, I wrote about my confusion regarding Weird Spotify Compression Artifacts. It turns out the artifacts I was hearing are not due to compression, but a result of audio watermarks that Universal Music Group embeds in digitally distributed tracks. This watermark is embedded in UMG tracks on Rdio, Spotify, iTunes, Amazon, and others. The watermark can also be heard in Universal tracks broadcast over FM radio. Universal Music recordings make up about 25% of most online catalogs, and its labels include Interscope, The Island Def Jam, Universal Republic, Verve, GRP, Impulse!, Decca, Deutsche Grammophon, Geffen, etc.
What the watermark sounds like
UMG uses a spread spectrum watermark, a technique explained in detail in this Microsoft research paper. The watermark scheme modulates the total energy in two different bands, 1khz to 2.3 khz and 2.3 to 3.6 khz. The energy is concentrated in the most perceptually sensitive frequencies because that makes it more difficult to attack or remove without significant audible distortion.
The energy is increased or reduced in 0.04 second blocks. The result can be characterized as a fluttering, tremolo sound. Listen closely to the original vs. watermarked audio samples and try to focus on the 1 khz to 3.6 khz noise range. It helps to wear headphones in a quiet environment.
Here is a short sample (excerpt: Three Doors Down - When You're Young). These are lossless original and watermarked files; what you hear is not a result of compression.
If the difference between the two isn't clear, here it is by itself:
The character of the watermark may seem subtle during this short sample, but through the duration of an entire song it becomes more familiar and more annoying. Check out my original post on the subject for more examples.
The watermark does not start until 1 second into the audio. After this the signal is divided into 0.08 second blocks. Each block is divided in two: some amount of energy is added to the first half and the same amount is subtracted from the second half. This coding scheme allows blind detection (without access to the original file). The actual information in the watermark is not easily recovered because it is modulated by a pseudo random sequence, which is generated by a secret key.
I did a little searching and it seems this watermarking technology is provided by MarkAny, a Korean company that has developed their own watermarks out of university research, and purchased some watermarking patents from Digimark.
Removing the watermark
Since the watermark creates audible distortion, it's worthwhile to try to reduce it. I wrote a MATLAB script that analyzes the block energy and applies some smoothing. This is the result:
Why do labels watermark tracks? Watermarking simplifies copyright enforcement by letting a company track music on peer-to-peer networks. "It gives them the ability to put pressure on policy makers and ISPs to do filtering," says Fred Von Lohmann, an Electronic Frontier Foundation attorney. That may be about the best explanation you will find. See DRM Is Dead, But Watermarks Rise From Its Ashes
Most listeners probably don't notice the watermark in the music they listen to. The audibility of the watermark is very content-dependent; classical, and solo piano in particular, are affected most severely. I've seen complaints on classical music forums with descriptions calling out the characteristic fluttering of the watermark. These listeners might be acutely aware of these sound quality problems, but blame lossy compression or streaming services.
I don't think there is anything inherently wrong with watermarking. The problem is with this particular poorly tested implementation. It is unfortunate considering the amount of engineering effort that goes into every music production.