Can you list them please. I'd love to abx them myself
The most famous ones are eig_essence, trumpets,castanets. You can get some of them from
http://web.archive.org/web/20070702152301/http://www.pcabx.com/training/index.htm I don't know where I picked up eig_essence but I remember it was linked to on hydrogen audio. Those are publicly available samples from real music that make some aspects of codec comparison/assessment very easy.
In my own collection i.e. stuff from CDs I encoded, I've most often run into clipping as being the main artefact problem (because some CDs are mastered close to 0dB and compression raises the level again), but occasionally also the kind of hissing you can easily get in the famous trumpets sample. Another issue I noticed sometimes was degradation of the illusion of width/soundstage.
I can offer some examples. I don't have an exhaustive list because I no longer have a need to abx stuff but some experiences stick in the mind quite well.
The most easily demonstrated clipping I encountered in lossy is in EMI CD 7243 8 26525 2 6 "Frühe Englische Orgelmusik" track 3. Parts are very close to 0dB and most lossy encoders take it over the edge, but not all. Vorbis did, opus did, lame did until the most recent versions of lame. I didn't use Quicktime or Apple but the frauenhofer aac doesn't clip it but attenuates, whether on encode or decode I don't recall. This track can induce plainly heard and very ugly clicking noises. Other people (lossy developers or proponents) have told me in one case that they can barely hear the clicks and doubt their existence, and in another case that I should have attenuated the level before encoding. Ho hum.
I came across a horrible hising sound in lossy encoding of Respighi's Violin Concerto in A Major on Naxos 8.572332. Off the top of my head I don't recall which track but it is probably track 3, the second part of the main work, which ends on a very long sustained high violin note which falls and rises quickly a couple of times. There is also some ambient noise from the orchestra (probably sheet music being turned over, breath being drawn and so on). Lossy encoding introduces some unfortunate hissing and makes a kind of mess of the ambient noise. This was with Fraunhofer aac and I don't remember if I tried other codecs.
One sample I noticed was on Jimi Hendrix's Bleedeing Heart from the Japanese Polydor issue of War Heroes, disc P20P22010. The first few seconds has a studio made effect. I guess the sound started off as cymbal splash and the groovy long haired peaceniks in the mastering studio transformed it into a very nice kind of whooshing sound which starts on the far left, moves left to right in an arc (the sound appears to travel up as well as across) and fades out to the extreme right. I was actually abxing the track for completely different reasons: I just thought it didn't sound as good as it used to (I'm middle aged so I feel that way about most things) so I decided to try to abx it. My subjective impression in normal listening was that the bass tones were less substantial or weighty and that was what I would be able to distinguish in an abx comparison. I couldn't, but while failing to do so I gradually noticed a difference in the "whoosh": the far left was less distant and the far right also. The arc of "travel" was less well defined. I did abx that and then my brain felt it might cave in on itself with the stress.
I found this really interesting because it showed me how hard it ican be to know exactly
why something doesn't sound like it ought to. There was no gross artefact, intrusive sound, nothing announcing itself. The differences identifiable in abx were small and it was actually difficult and somewhat exhausting to do the abx. The single thing I could abx was something I hadn't even considered, yet in normal listening I had realised that I wasn't hearing what I ought to be hearing without going to any special effort beyond wanting to enjoy the music.
If that hadn't been a track I had heard many many times over the years then I probably would never have noticed the difference. It also still sounded like its bass quality wasn't right in normal listening, even after I had abx'd it and not been able to identify a difference in bass. That may be a problem of my echoic memory, poor interpretation of sound on my part, not having adequate terminology/expertise/experience to make a better diagnosis. It may be that subconscious bias is so powerful that it trumps knowledge, experience and practise even while subconscious bias is being consciously considered, I'm not sure. But that conundrum was solved entirely by listening to the flac instead.
