I was curious about this phenomena when I first heard about it. Interested..but not so interested that I had to rush out and find things out about it right away.
Recently, someone on the forum mentioned it again. Then, I remembered, about 1 year ago or so, I was out at the local version of the Sally Ann,or Goodwill store. They resell used goods for charity-I cruise for neat audio gear, but it is seldom found these days. I had found a rat-shack bulk demagnetizer (new in the box), from the days of yore, when hackers really had to erase their shit, when they are telling the coppers to wait "just one more second!" before they get to the door to let them in.
I heard this recent bit here on the forum about demagnetizing and it got me thinking again. But I could not find the damn demagnetizer. Yesterday I was looking desperately for something else and stumbled across the demagnetizer I bought that year ago.
To make a long story short, I did an experiment where I played a bit of two different CD's so far, and stood in the exact same place in each instance, both before and after the demagnetizing.
My expectations where neutral, I had no idea what to expect and don't recall exactly what people say they hear when it is done. This means I was trying to start with as little bias as possible.
I used Tracy Chapman's first album and The Mavericks: 'Music for All Occasions' for this test.
First I played about 30 seconds of the first three cuts on each CD.
Then, I demagnetized each for a solid 20 seconds in the area of the demagnetizer that concentrates the greatest amount of magnetic AC force or energy, this..by moving the CD's surface around on the demagnetizer in slow circles to get the given CD's metal layer through that area as much as possible and as evenly as possible. Slowly, as well, not fast. Similar to proper demagnetizing technique, just for the sake of using a known technique, even though the application and result is likely to be different.
What I experienced on playback was a drop in the intensity of leading edge transient characteristics and then this effect headed into the treble, which is principally all transient in nature.
What this imparted to the signal reproduction was a seeming reduction in transient hash, softer dynamics, a slightly more open acoustic and a slightly more warm sound. Rhythmic drive was as stable as before, but seemingly softer in 'hardness' of overall drive (hashy edges where lessened in value or intensity). In the Mavericks CD, a bit of violin in the usual slightly screechy country kind of emphasis and way is used, and the emphasis was slightly less 'harsh' or less 'edgy' or less edging into break-up. Ie, reduction of a tendency to go just past an acceptable edginess that a good recording engineer would only reasonably attempt to do/capture (as it real), but CD and digital playback can and do exacerbate. Ie, the tenancy of digitally recorded and played back horns to rip you to pieces like shouted and heaved broken glass, where a more natural effect is to be resonant and have severe drive and 'thru-you' intensity, like a real horn does. After demagnetizing, this digital artifact was ameliorated to a small degree. Notably better than before.
This overall result brought me to one point which is the likely culprit insofar as explanations go.
Pit Edging, which defines the beginning and stop point of the laser when it comes to defining a 'pit' or 'bit' on the CD.
I feel the very leading and lagging edges of the pits have been likely 'rounded' by the intense electromagnetic field off of the demagnetizer, and thus the 'pit definition' is more perfected, and thus, due to how the signal is pulled off the CD's, there is a slight reduction in Jitter borne effects when it comes to bit placement (meaning start-stop bit/pit definition) in the read process.
a secondary component of getting the start-stop of a read to be more defined, with respects to pit edge definition is 'laser splash' inside the CD's plastic polycarbonate layer itself. when the laser hits the edge of the pit, there is a short time where the laser light is spread out and reflected all over the inside of of the disc plastic layer, in the horizontal domain. There is a possibility, however low, that the S/N ratio, at the point where a bit is recognized as existing (beginning)and then recognized as ending is slightly more perfected as 'light splash' may be reduced by this demagnetizing effect. It IS possible to surround the periphery of a CD with LED's and make a CD sound absolutely terrible by screwing around with the read quality in these sorts of ways. This consideration has been known as long as the 'green pen' on the edge of the CD disc days have been around. The green pen is a legitimate thing to do for that reason alone but is actually more subtle than demagnetizing, in my experience.
A someone who plays with the innards of CD players to an extensive degree, I do know my way around a clock mechanism (and every other possible chip and component, chassis, drive mechanism, etc) and know what jitter issues sound like under all possible levels and circumstances. I do a minimum of 250+ single cause analysis experiments in audio every year, and for over 20 years, this has been my daily fare. So I DO know what I'm speaking on and about, as subjects go.
There are very high odds that I will demagnetize all CD's I own, but I have to be sure, of course, that it is a safe effect to pursue with respects to CD lifespan.