One thing I wonder about - do all CD-ROM Drives / operating systems preserve polarity for an audio rip or could some of them be flipping the sign on the integer sample values that end up encoded in the WAV file? Did you rip all these on the same drive / OS with the same software and load them into the same DAW? On my system it sounds like the '83 (ripped and de-emphasised) is in the same polarity as the 09 (CD rip and USB files) whereas the '87 is the odd one out - will listen again. Even the de-emphasis program could invert the polarity if not coded carefully.
Some drives had issues swapping channels, but in 20 years I've never heard of a drive inverting the signal.
Yeah - I had one of those So, have you tried ABXing the "Something" samples with both in the same polarity? The difference you were describing hearing seems to match what I hear when comparing the two polarities on just one mastering - aside from soundstage difference there's a sense of different EQ although no such adjustment has been made.
AKA The Wood Effect: cjwoodeffect » I suspect it may have to do with an asymmetry in the elastic nature of a fluid (air in this case) as the pressure wave passes through but the complexity of a real world example is so enormous that it would not be easy to show from theoretical arguments. Asymmetry can be seen in the recorded waveform, however.
Okay I'm not a physicist but I know this. If you have two waveforms which are are exactly the same with the exception of having opposite polarities, they'll cancel each other out. Means that you don't hear anything anymore whilst you play back these two files at the same time.
Any asymmetry in the waveform will be mirrored across the time axis when you invert so you will still get cancellation when added together with the original waveform. However, in inverted playback, stretches in the elastic fluid medium (air in this case) that were recorded will be presented as compressions and vice versa, a difference which can be detected by the human hearing system.
I'm glad Mal can explain it. I can't. All I know is I can tell the difference with my ears. Some people seem to notice it, some don't.
No chance of this, really. I rip everything twice with two different drives using two different rippers (EAC and CUERipper) and verify the two rips are identical. This is certainly overkill to verify accurate rips, but is my habit. The software for comparing waveforms is something I wrote myself and is not negating the integer values. My hearing is not good enough to corroborate. But wouldn't mastering differences render pointless any attempt to hear polarity reversal? I must admit to not owning the actual 1983 CD, but have a "copy" which I presume has the pre-emphasis intact.
I think it's possible to narrow it down to the assumption that fluid compression/rarefaction are equal and opposite is flawed - i.e., the restoring force function is not identical in form for both when you have a wave travelling through a fluid. How you show that from a theoretical standpoint is not so easy to explain. It involves getting down to some pretty heavy subject matter but I think it can be understood from a more general point of view by referring to the concept of complex numbers. If you consider '1' to be the reference then in the sense in which we are talking, 'i' is its inverted form. It's not the same as just saying '-1' is its inverted form. While the latter interpretation may appear to be true for a time-domain representation of a waveform (inverting simply requires changing the sign of all the samples), it is not true and doesn't even seem to make sense for the frequency-domain representation which must be considered as equally important as the time-domain as to what constitutes the information in the waveform. Sounds a bit strange, I know but we live in a Universe governed by the rules of complex numbers so maybe it is not so crazy after all. Think of it like this - suppose you reflect an image using a mirror - everything appears reversed relative to the original image, but there's more too it than that. The new image is 'in' the mirror... I think it's possible not to notice it - but it's there, unequivocally. Yeah, just wondering if occasionally there's some code out there that inadvertently flips the sign, inverting the polarity. I'm sure I've heard of a DAW that did this? It could get in the way in a spot check, certainly. However, I think with prolonged listening it would become clear. For example, I'm hearing the vocals set further back when they sound "sweeter" on the CP35 - inverting seems to bring them out in front but with less definition so less convincingly away from the plane of the speakers. That's what I am going to listen for next on the 09s. So rather than rely on the 'apparent EQ' (unchanged but seems slightly different when inverted - perhaps better to call it timbre?), I think focussing on the soundstage should help identify inversion independent of mastering differences.
Listening again, I'm thinking the '09 24-bit FLACS sound even better (edits notwithstanding) inverted (confirmation bias? I don't think so but you don't have to take my word for it). My take is that the CP35-3016 is rated highly by some because it presents the vocals along with other key elements in absolute polarity (I would guess the majority of playback systems are also in absolute polarity). In other words, all the other digital versions are improved (to my ears) by inverting them, if you prefer the less 'pinched' and more set back vocal sound (I'm finding improvement in most other key elements too with nothing sounding off so I'm definitely preferring it inverted with respect to the standard digital mastering polarity). It's worth noting that the '09s sounded better than the '87 without inverting them and were more similar to the CP35. My guess is that the clearer top end and increased resolution of the '09s compensated to some extent. Upon inversion the '87 and '09s improve but the '87 remains behind the '09 (24-bit - not so impressed with the CD, again just my opinion in case any hi-res deniers are listening) whereas it becomes closer to the CP35 (which falls behind the 09 24-bit except for the fact that it is unedited!). I would have to spend a bit more time listening to the CP35 and the inverted '87 to figure out which is better but for now, I'm just enjoying Abbey Road the 'right' way around I suppose you could argue that there is no right or wrong here, or even that I'm listening to it 'wrong' if it was mastered with the opposite polarity in '87 and '09. However, vocals are sometimes used as the reference for determining optimum playback when other parts might not all be in the same polarity. YMMV.
Would love to hear one of these without massive amounts of groove distortion and crackle on nearly every track though. Makes me tend to opt for the BT CD or the -6/-4 LP.
Thanks Mal for the breakdown, one of these days i'll do an invert on the 87 and see how she goes Japan Duran Duran Rio has incorrect polarity, but I noticed no difference once corrected
It's because of the hair cells in the human inner ear. The neurotransmitters fire off when the hairs are bending one way, but not the other, so instead of receiving both the peaks and the troughs of the waveform, your brain is receiving one and using it to reproduce the other. If the wave is very asymmetric, that reproduction won't go quite right. (I'm oversimplifying somewhat. There's a more detailed discussion near the end of this PDF: http://www.phon.ucl.ac.uk/courses/spsci/AUDL4007/12.pdf) Of course audio hardware might also be an issue, e.g., if your speaker cone has more distortion moving forward than backward. In practice, I don't normally notice phase inversion. I can usually distinguish it in blind tests when listening to something with a very asymmetric waveform, like drums or brass, but it's really, really subtle. It's best to check it out yourself with an audio editor if you're curious.
Um, I'll give it a try .... The inner ear is where sound vibrations get converted into nerve signals. The cochlea has lots of cells with tiny "hairs" (cilia). The cilia move back and forth with vibrations, and sometimes they fire off nerve signals. But they're much more likely to do this when they're bending in one particular direction. Here's how a nerve fibre responds to a 300 Hz sound signal: Get a whole bunch of cilia doing the same thing, and your brain gets an image of the pattern of the wave. Or rather, half the wave ― the troughs, not the peaks (assuming the text I took this from is synching the two images up correctly). For a pure tone like this, that doesn't matter. The troughs are the same shape as the peaks. If we know what the bottom half of the waveform looks like, we know the top half. But if the waveform has an unbalanced, irregular shape, the signal from the top half of the wave can be quite different from the signal from the bottom half. Play a sound and then play its phase inversion, and our brains will receive them in different ways. Here's a trumpet waveform: Flip this upside-down, and it won't sound the same. It'll sound very close! Our brains are good at doing the math required to fill in the information they're missing. But it won't quite be identical.
IIRC, the early Virgin Dire Straits BIA CD also has incorrect polarity. Never gave it a thought to listen with the correct polarity, might have to give it a try.
the 1978 reissue is nearly just as good. Minty ones go for 40-50 dollars. Less than half of what you have to shell out for the japanese cd
As requested: Come Together sample 1 [FLAC] Come Together sample 2 [FLAC] Once again, no processing except lowering the level of the 1987 CD by 2.97dB to match the Odeon. I'm a little tired, and I haven't listened as much as I did to Something, but I'm still finding the difference to be extremely minimal. The biggest difference I hear, which I noticed before, is the balance between the two is slightly off, with one pulling more to the left and one pulling more to the right. It's subtle, but you can detect it when switching back and forth. Not sure which is correct.
Do you know about AccurateRIP (www.accuraterip.com)? It can automatically compare checksums of CDs you rip to a database of checksums from other people's rips of the same CDs. No need for you to rip twice.
The talk on polarity here reminds me of this post: The Beatles Remasters: Abbey Road » I'd guess that fixing the azimuth error on the 1983 CD may make more audible difference than correcting polarity. I've never got round to trying it though.
Yes I know, thanks. In addition to AccurateRIP, CUERipper also verifies with it's online CUETools database, and may even be able to correct remaining errors in the rip. I'm not recommending ripping every CD twice, or suggesting it's necessary for verifying an accurate rip. It's just my own, slightly OCD habit. It's not that much extra effort really, as I usually rip more than one CD at the same time, and can usually keep two drives running simultaneously.
Thanks for that. Yeah I'm convinced, listening through my headphones I can barely tell the difference between these two files. Is sample 1 the black triangle?
