I thought ok, I'll rip a couple of these having trouble to Song Wav files and use No cue file. They still play back with crackles. I'm not sure I've mentioned this? So small individual song wav files with no cue files around are exibiting poor crackling sound.
Dave, please stop looking for more albums that exhibit this; better start listening to those who don't exhibit it. BUT :
I have no technical explanation for this apart from it being real. But I guess that is stupid thinking of me ...
I must say honestly that I forgot to look into the fact that the Crack Detect shows nothing at the first track, but does at the second, while it should be so that the complete album is scanned in either occasion. This stinks. This suits your observation about the first track not crackling (at least you did not mention that I think) and the second does. But how in the world can this be correlated to the individual tracks show the crackles as well ?
I have a major stupid idea again :
Let me first tell you how surprised I was that the transients can be as high as I found. So, it is the transients I use for checking for crackles, and believe it or not, it trips when over 20 times a decimal value change of 60000 (!!) has been found (while max can be 65536 for 16 bit data).
Note that I started with a value of 100 to rip upon, because I thought that would do fine. I then ended up with 8000 which could occur over 50 times on one track, and when I found this, I thought I had some news for this world, or IOW I thought now all CD recordings s*ck all over.
Then I asked a recording guy who could know, and he said that transients of close to the full 65536 just could happen. So, I was satisfied with that, and built in the tripping at 60000.
Btw, from pure electrical thinking I can't believe this is right, because a negative voltage near its far end, goes right to positive to near that far end, and that happens within one adjacent sample. Whatever instrument or voice could do that within 0.00002 second ? But the guy said it ...
At this moment I can't tell anymore how many things are wrong at audio playback, but I do know that things can be improved. We
claim this is jitter doing that, because we don't know other explanations, but please keep in kind : I already think for a longer time it is not jitter what we're influencing. Or not jitter *only*. Ok, just keep this in mind, in case it becomes important later ...
When I found those huse amounts of decimal 8000 transients, I looked at such a file for confirmation. Btw, the 50000 occasions of decimal 8000 in one track (the 8000 being an 1/8 of the total range) is a relative small amount looking at the number of samples in a whole track. Mind you, there are 44100 in one second, so a 300 second track would contain those steep transients in around 1/300 part of it, totalling to one second, or roughly 150 occasions per second (never mind my math).
How can it be so that 150 times per second a transient occurs of which at least I myself cannot imagine how it can happen in nature ?So, looking at the file - and I often look at files - it indeed showed those "endless" going up (or down) straight voltage changes, but, I never saw them before. Why ? well, because when there are 150 in a second, there are 43950 more normal sample-changes, and the chance to meet a strange one is 0.3%. Also, if you are expeerienced on this, you will know that you'd *never* find such an occasion by accident, which is related to the scale you are looking at, and or the zooming is so small that the peaks won't show, or the zooming is so large that only, say, 10 decimal fits vertically in the window, while we talk about 8000.
Well, since the Crack Detect showed that they are there, and aith some addtional data I knew where to expect the "cracks", I indeed found them. It looks completely ridiculous, and I keep on saying it can't be. On this matter, note that any vibrating source needs a start for a couple of cycles before it comes to full amplitude. Even electrically (like in an amp) this is so, I think.
Rather important side note :When indeed it is normal that such huge transients exist, upsampling just can not work, meaning :
When a transient with 16 bits comprises of 8000 decimal, you tell me where to split that transient. At 4000 ? at 7000 ? at 100 ? mind you, it happens at 4000 because it is the only thing to "know". This is *very* different when one expect transients so be, say, 20 at max, and the relative error would be infinitly smaller. Also, who says that such a *transient* should be cut ?
Back to my rather pretentious thinking I tried to start this subject with : what if playback implied so much more accuracy (of whatever it is that can happen in the DAC) that we start to hear those "crackles" ? And please keep in mind : those huge transients MUST be heard as crackles. I wouldn't know how to perceive them differently. And they just *are* in the data ...
To put things in the proper perspective : when one such a transient would be there I think it is virtually impossible to hear it. On that matter : 0.00002 seconds is the time one sample takes, and the crackle you hear sure takes longer. This
could be caused by nature physics (like ringing), but I don't think so. What I do think, is that only more of them can exhibit as being there, and indeed in the file it's not single spikes showing. It's always more, and it always looks like no accident. So, say that 10 of those become audible, it's only a matter of meeting them at the right place (like in the otherwise coincidentally low amplitude of ambient sound).
Dave, so far I only listened to the second track of that Cornershop album, and as I said before, this didn't sound right to me, *but* it was on the edges of the rythm. Wat does this tell ? bad recording ? something overhoots ?
In order to understand the difference between versions ... things *have* changed, but if all is right, nothing that could incur for something being wrong now. More accurate (in my theoretical perception on things that change sound) yes. The last version (0.9u-8) more than the one before.
I too have been walking to the speaker a couple of times because from a distance I thought I heard a "distortion", but being upclose I thought
.
All of the above may later turn out to be somepletely senseless, when I find something wrong in the software afterall. But if the latter is not the case, the world news is written in the above : CD recordings s*ck all over, and it only needs more accurate playback to discover it.
However, so far I believe the guy who told me that those transients can exist. But keep in mind : this is what he saw on his DAW, looking at the master before pressing to CD. So in the end he might just have seen the same as I did.
Another weird thing I ran into, a few times (like once/twice a listening session) 09-u7/8 would ... like... the musicians would false start but immediately start up again from hiccup, this was usually at the beginning of a song. no distortion, kinda seemless. If you're listening to rock it might seem in place
I believe it, but can't imagine what you exactly mean.
