CD's are not lossless. Pretty good though.
Is an audio signal lossless? That question doesn't make sense. Analog audio is limited by bandwidth and signal-to-noise. Digital audio is limited by its Shannon-Nyquist parameters.
(Monty's show and tell for geeks is great if you have even the slightest interest in these things!)
Vinyls by definition is very lossy format (low dynamic range (although there was obscure "HD" vinyl format once)).
The science is here: https://xiph.org/~xiphmont/demo/neil-young.html
Edit: You're being downvoted which I think is unfair. It's completely logical that higher sample rates and higher bit depths "should" sound better, and certainly the music industry wants you to believe that. Unfortunately it's simply not true above a certain threshold.
"Lossless" here is defined as "stays the same once it has entered the digital domain", and not "identical to the real-life source", and so CDs are in fact lossless.
That only applies if you have a perfect master.
If you want to listen to a song mastered for CD, unless it’s 24/192, it’ll likely clip at several points.
Which is about 90% of songs today.
And if you want to remix the song, or use an equalizer, or other effects, you’ll also want 24/192.
And nowadays the ultrasonic effects he describes don’t even affect mobile phones anymore, and storage is plenty – making the disadvantages he describes moot.
There’s thousands of retail CDs out there with clipping, intentionally created during the Loudness War.
To listen to them properly, one would require a remaster, or 24/192.
Sadly you have no guarantee that a random 24/192 recording wont have the same compression / clipping problems that plague CDs thanks to the loudness wars.
Which is quite awesome, because that doesn’t have clipping.
Basically, what I want, is ideally the master, in the original version.
Just like I’d prefer RAW images with a header telling my program which settings to apply for rendering – but still allowing me to change exposure.
That's certainly not the usual sense of "lossless" as applied to recordings, since it would make mp3s lossless as well. We generally say that mp3s are lossy because the mp3 encoding process loses information that was present in its input, not because the encoded mp3 degrades when played (it doesn't!) or when copied (it doesn't!).
CDs, as the original source of music, cannot have this sense of "lossy" or "lossless" usefully applied to them. (You could claim that the performance is the original source, in which case it's still not possible to apply the lossy-vs-lossless concept, as no lossless recording of a physical performance is possible.)
If we think thoroughly, one can say that anything that comes through the cables, amplifiers and speakers is lossy. Just because it's not the direct transmission of the sound from the music instrument to a human ear through air. As we all know, our ear is a pretty complex audio sensor that consumes sound waves from multiple directions (perhaps, the unlimited number of directions) - direct and reflected waves. So if someone says that all the performances of every rock or electronic band are "lossy" just because the sound gets transmitted through a limited number of speakers (not to mention the number of sound effects getting applied - distortion, hi/lo-frequency filters, reverb, compression etc.), that guy could be onto something in actual fact. That position also has some grounds and rights to exist.
Also, I doubt one can precisely emulate the actual performance of an artist with just 2 (5, or even more) recorded channels. As every performance, from the listener perspective, also depends on the location, current place's acoustic parameters, etc. (say, if you are in a garage with metal walls, the sound is reflected differently than in an open field or mountains). As well as different people attending a music event will hear different kind of sound, simply because each of them will be located in a different position and receive different/unique sound waves from different directions. From this perspective any record of that exact concert would be "lossy" (not the same) for every single person.
Some people in the thread discussed whether the audio record keeps lossless after mastering. Well, you could say, applying any filter to a record (regardless of whether your intention is to make it sound better or not) distorts/changes the record anyway. Especially keeping in mind that it may involve eq-filters and compression for different frequency bands. I mean, do we lose something in that case? Sure. Do we call it "lossy"? I suppose most people don't.
The other guys tried to use the limitation of a human ear as an argument as well. But this is also debatable. Where would we stop that way? Exclude all the frequencies above 22KHz off the consideration while speaking of "lossless" vs. "lossy"?
I might be mistaken here, but I think applying the term "lossless" to CD quality is a convention amongst the most people simply to have some kind of a common measurement standard to refer to as a basic "unhurt" record in the digital space. Whether it is correct or not technically... Well, I am not sure it really matters now.
PS That being said, you seem to have got down-voted wrongly. While you may have an opinion which is a bit different from what the other people have regarding the use of "lossless" thingy, you don't seem to have said anything bad. I am voting you up myself just trying to negate that wrongdoing. :) BTW, I even didn't know that one can down-vote here. Every time I visit this web-site, I feel like I got back in 80s in terms of how this forum looks like (UI/UX). I wonder when YC will finally do something about that and align it to the contemporary web standards. It's super weird that the company funding most promising startups in 2016 has a forum-like kind of stuff lost 20 years behind. Sorry for the off-topic.