Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Apr 29, 2012 5:34:46 GMT
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Apr 29, 2012 8:31:52 GMT
Although a fun experiment it still has it's pitfalls. It only proves something about the general concensus (with near 50%, the same as chance if enough people experiment) and not about the individual taking the test. Same as the Meyer and Moran test b.t.w. It doesn't 'filter' out the people with the better hearing as they are dropped on the same pile as those that fail. What's my reasoning for this... 2 files that have minute differences (proven later on it does) caused by added dither. Now here is the sting in the tale. You listen back and forth and either pick one as better or not. Too bad Alex didn't pick one openly, like all the other posters did, and only felt the need to describe both recordings as poor MP3's and went out on a limb to support the hires cause. (just a tease and your every right to post Alex...) Now the results of the Italian job: Identical = 11; non-Identical = 20. Preference was almost equal between A and B. what can be the overall conclusion ? 31 participants (some capable of analysing the differences by PC) with 31 persons it gives a big enough sample to tell 'gamble or not' to statistically minded folks. About 30% couldn't tell. About 30% said 'A' about 30% said 'B' Well that appears as 100% gambling... but is it ? on CA: 35% couldn't tell. 30% said 'A' 35% said 'B' For instance PeterSt first 'gambled' 'B' but after a while (while differences were revealed slowly and perhaps looked at the files, like Alex did) changed his mind to a definite 'A'. Fair ? perhaps, perhaps not. So of the difference people 35% preferred the dithered sound, 30% the original BUT 65% were of the opinion they could tell. Looking at chance most of them preferred the dithered signal which is otherwise the same and is 'close enough' to 50% given the sample size (9). 30% picked the original as the better sounding one. Does this prove you can discern between a file with added dither or not ? statistically it proves it isn't possible. However, on a personal level 30% picked the undithered as preferred and thus preferred undithered making them 'right' BUT also those that preferred the dithered signal are 'right', they just preferred the dithered file. NOTE: Dither is an ANALOG noise (+/- 1LSB in level) added for reasons to improve fidelity of low level low and medium frequencies DURING the sampling process. Afterwards it is only adding noise really and not 'contributing'. Anyway... 35% preferred the added noise which is at 1 LSB level at 24 bits (around -145dB ?). Since this was an LP rip we can assume the hiss from the album+preamp is around 60dB and would be 'masking' as well. So to those that did choose between the files (better hearing capabilities) the score = 100% they can discern (they picked) and 44% 'won' and identified the 'original file'. definite proof they CAN hear differences and also definite proof preference plays a role. Furthermore it is also proven some people (35%) cannot hear it as they expressed they couldn't discern. gears or ears ? Final conclusion: hearing people (clearly ?) heard differences and proved that. Statisics had proven the odds are close to 100% chance and on average humans couldn't tell. This is where the 'test' isn't very valuable. a '2 bob each way' test. IF this test would have had to have more 'proving power' there would have needed to be: a: someone looking over a shoulder they didn't cheat by examining the file. b: for statistical reasons at least 10 files should have been uploaded (with one known reference or not) and the correct ones had to be picked. I say .... this test ONLY proves something to the individual taking the test 10 times or more without 'looking' (ABx in Foobar for instance) The test being what it is now proves ... well... nothing to either skeptics or believers and both will find definite confirmation of their 'religion'. ;D This is why I am all for (blind) testing ... but ONLY when done the RIGHT way.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Apr 29, 2012 9:02:12 GMT
I didn't look at the first 2 files. I listened to them, and thought B may have sounded a little better. However, I decided that the quality of these 2 files was quite poor for 24/96 files, and of insufficient quality to pass a definite judgement. I deleted the files days ago. I thought those later ones linked to by Peter were slightly better than the previous and tentatively selected A. Again, I thought that the SQ was nowhere near good enough to reliably pin point differences, and only today checked the 2nd lot using SF9. Both file A and File B show that same low level pattern in SF9. Attachments:
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Apr 29, 2012 9:12:21 GMT
TRACK B. P.S. It's a shame they couldn't find something much better than digitised vinyl . Attachments:
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Apr 29, 2012 9:32:36 GMT
I agree that this individual had better used a GOOD quality 24/96 file and altered that. Why choose a vinyl rip in the first place.
But as said... this test proves nothing to nobody.
A waste of time to conduct ill conceived tests when in the meantime you can enjoy music. If you want to test... do it right.
In the EMC world, where everything depends on proving to an independent and notified body something meets certain criterea while the demands and rules are clearly outlined, there are still ways to 'cheat' to get approval for equipment. Sometimes the actual marketed product even differs from that what was tested.
This only becomes a problem when something VERY bad has happened (people killed, severly injured or HUGE amounts of money (claims) are involved AND in a court of law that manufacturer is found 'guilty'.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Apr 29, 2012 9:37:29 GMT
It may have been more meaningful if they had also posted the file before it was stuffed around with.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Apr 29, 2012 10:48:24 GMT
Perhaps I need to have a shower and think of going to bed early, but doesn't this mean that the original File "A has not been altered, and we are comparing the original untouched file "A" with a new version, file "B", which is simply File "A" with triangular dither added ? I am now thoroughly confused because both the downloaded Files "A'" and "B",(posted earlier) both show the effects of the triangular dither in SF9. I have this evening DL the original files again as at the initial link, and the results are the same. Will someone please explain to this tired old brain what I am missing here ? Alex
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Apr 29, 2012 13:18:20 GMT
I have now posted in C.A.that given that all 4 files look the same in that area in SF9, that it's no wonder that results were random ! It still smacks of a conspiracy to me, or perhaps , to be more charitable that it is due to an error duplicated in both the rar DL and the 2 wav files DLs. Either that or I have made a monumental stuff up, even by my standards. Alex
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Apr 29, 2012 13:22:32 GMT
You are missing the point.... being... the test is quite meaningless and a waste of time. It adds nor removes anything from the 'good old' debate. I for one have not wasted any time on this except for giving my opinion. A-B ing this 10 times and keeping scores will very likely also end in a tie (between 30 and 70%). Would be amazed if someone could score 9 or 10 out of 10 (blind, so using ABx from foobar) hmmm.. seems like that Julf character (I thought he was banned ?) has the same opinion as I have.... the test being pointless.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Apr 29, 2012 13:32:40 GMT
You are missing the point.... being... the test is quite meaningless and a waste of time. It adds nor removes anything from the 'good old' debate. I for one have not wasted any time on this except for giving my opinion. A-B ing this 10 times and keeping scores will very likely also end in a tie (between 30 and 70%). Would be amazed if someone could score 9 or 10 out of 10 (blind, so using ABx from foobar) Frans The OP in C.A. has now confirmed that file "A "remained untouched. Yet ALL files show the same as in my posted Screen Grabs. That is plain bloody wrong ! Perhaps it was all started in Italy on April 1 ? Alex Julf wrote a long email to Chris Connaker and was reinstated as a member. He has been much more diplomatic since returning.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Apr 29, 2012 13:57:08 GMT
When the files are indeed the same (some don't think so and IMO can only be seen in nulling tests not by looking at waveforms) then the test has had great proving powers namely 4 identical files where people express a preference that isn't there. Somehow the added noise (around -145 dB ?) would only be 1LSB high and random. The amplifier and soundcard would also need to be able to reach those levels as would your ears. Playing at a deafening level (say 130 dB at full power above the pain treshold) that added noise would still be 10 dB BELOW the hearing treshold.. that's at least twice as soft as undetectable already. Remember deafening loud as in close to speakers at a pop concert while the background noise would be so low it wouldn't even be heard in the quietest room in the middle of the night. In those quiet conditions suddenly play music of 90dB... you would feel as if your deafened... now add 50dB on top of that to get a feel of the levels we are talking about. The noise levels of the recording (vinyl playback) are also at least 5,000 x higher than that of the added 'dither'. pointless exercise. Similar to reading this nonsense: people.xiph.org/~xiphmont/demo/neil-young.html
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Apr 29, 2012 22:03:14 GMT
Vinyl has been disputed, yet wow and flutter can be clearly seen when scrolling through in SF9.Perhaps from a tape deck in that case? With a tape deck the noise floor is likely to be around -60dB, and perhaps a little over -75dB with noise reduction. Claims have also been made that the level of noise shown is normal. As Frans has said, the whole test is a waste of time, yet some who should know better are still in denial.
|
|
|
Post by dalethorn on Apr 30, 2012 1:09:11 GMT
I read the entire XIPH article, which most of it is compelling. The strongest arguments for 16 bit / 44 khz being good enough for perfect reproduction forever seem suspect. There are too many requirements - too many things to go wrong. I agree that allowing ultra wide bandwidth can introduce problems, but that doesn't seem much different from having to deal with turntable rumble - get better gear instead of just chopping the sound below 20 hz. I can't dispute the claim about digital making perfect copy of music waveforms that fall within the limits the author defines - but it still seems unlikely. The author admits to subliminal perceptions of loudness at 1/5 of one DB difference, then basically denies any other subliminal effects, or says they fall within the 16 bit / 44 khz sampling range. I would like this paper a lot better if the author just recited his known facts and test results and didn't jump to conclusions.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Apr 30, 2012 5:40:14 GMT
Some of the experiments I have done myself, like hearing with my own ears where my detection borders are, and had to build my own test equipment for it. I found 0.1dB to be inaudible (and that was sighted). Some couldn't tell -0.5dB apart (blind) though, blind I managed to pick out 0.2dB.
He has a test with 2 loud tones of 0dB 30kHz and 33kHz and shows how much garbage comes thru in the audible part with 0.1% distortion. I have NEVER seen any music files with contents this high in amplitude, more likely it is far below -40dB if there is any (a lot of recordings are chopped of above 20kHz for technical reasons) That 0.1% seems like an old and outdated figure.. and it is, but at the same time with SOME amps the distortion starts to rise rapidly above 10kHz. So an amp with a THD (@1khz) of 0.001% may well have over 1% of HD at 30kHz. Better High-end gear (aside from tube amps with trafos that distort heavily at these freq) is not likely to be bothered.
the -105dB is possible with dither. It works similar to 1 bit DAC's, they even manage 20+ bits resolution with just 1 bit) but whilst the -105dB tone can be heard (a single tone) you cannot clearly hear music (or lower level noise) anymore, the brain simply isn't capable detecting a wide band signal in wide band noise. Marconists were trained to detect even lower levels of 'beep' in heavily noisy signals. Most people couldn't even hear that beep. But if there were more than a few beeps they couldn't anymore.
There is a LOT to be said for having a 24bit DAC in the home. No denying but when not playing at very loud levels 24/44 cannot be distinguished (Meyer and Moran).
In my younger years where I could easily hear 18kHz and still felt 'pressure' with 19kHz there already was controversie about the limited bandwidth and 16 bits. I wanted to know for myself and I made an AB box with the possibility to insert a sharp cut-off analog filter in the chain. My phono was capable of reproducing over 20kHz and seen content present on a scope. I had (dynamic) speakers that could reach 40kHz and I (and some others) tried for a long time to see if we could tell. The talk of that time was not so much the limiting but the sonic aspects of the ringing of 2nd gen CD player filters. I used those.
Needless to say there was never a moment we ever heard not the smallest difference with or without filter. When you sweep a tone generator this isn't strange as well.
Having audible content below 20Hz can still be usefull when playing at high levels. You can't hear it with your ears but the sensation of vibration adds to the experience. I had a 11m long room and 16Hz-0dB sub and on one recording the whole floor shook. No internet then so didn't know WHY this happened but was in the (classical) recording. Turns out underneath that theater a subway caused this. Impressive and great add to the experience though not music related. Never heard that on a HP.
Anyway the only way to experience WHAT can and cannot be perceived (by yourself) is to test and experiment. I learned not to trust reviews and hifi mags for several reasons so experimented and was quite enlightning. we only had 16/44 in those days.... The rest (what the borders are) isn't any different from those days.
|
|
|
Post by dalethorn on Apr 30, 2012 14:43:00 GMT
I've listened to large pipe organs in churches so I have a good idea what subsonic music frequencies are like. Not much different from earthquake waves. And neither case has to do with the shaking caused by the sound waves. But the interesting thing is that I get the same sensations of sound with a good headphone as I get with large organ pipes or earthquake waves. And again, don't confuse the sound waves with the aftereffects (shaking, etc.)
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Apr 30, 2012 15:37:55 GMT
I had a wooden floor.. The whole floor actually moved up and down. When playing loud it actually became scary ! obvious the freq was (almost ?) the same as the floors resonance freq. moved an now I have a concrete floor and smaller room (7x7meters) so the lowest notes don't fit that well anymore. DT1350 is amazing on these notes though and because of room acoustics the DT1350 reaches down lower as my room does. Incredible... similar to LCD2/3 in that aspect. Those who say you can't hear really low notes (20 to 30Hz) and only feel them obviously doesn't have a headphone that reaches that deep. While the lowest organ note on most organs is 16.4Hz (32 foot pipe) an organ is capable of producing frequencies with a considerable lower frequency as well. There are organs with a 64 foot pipe reaching 8Hz. acc to a website: The lowest note is produced by combinding a stopped 64' ( 128') and stopped 42 2/3' (85 1/3') to produce a resultant 256' which is 2Hz on CCC CCC C, this note is not considered audible by the human ear.
|
|
|
Post by dalethorn on Apr 30, 2012 18:38:02 GMT
I have wondered for a long time how the DT-1350 goes that low, since it doesn't seem to have much mid or upper bass - not prominent certainly.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Apr 30, 2012 19:28:46 GMT
good seal, good membrane compliance.
It isn't loud or obvious (as in DT990, D2000 e.t.c.) but as flat as can be and runs below 20Hz. Most people don't like flat, I do as well as 'colored'. An aquired taste I reckon.
|
|
rowuk
Been here a while!
Pain in the ass, ex-patriot yank living in the land of sauerkraut
Posts: 1,011
|
Post by rowuk on May 2, 2012 5:22:39 GMT
It is a rare pipe organ with ANY type of 64 foot register. Most in Germany have stopped 16 foot to simulate a 32 foot register.
Subsonics in acoustic music consist of the standing wave in a room. THAT is the information that tells us how big the room really is. That standing wave CAN be in the recording (if good omni condenser mikes or ribbons were used) but is ALWAYS present in the listening room too.
External noise sources are usually not that big of a deal.
|
|
|
Post by dalethorn on May 2, 2012 15:37:43 GMT
It is a rare pipe organ with ANY type of 64 foot register. Most in Germany have stopped 16 foot to simulate a 32 foot register. Subsonics in acoustic music consist of the standing wave in a room. THAT is the information that tells us how big the room really is. That standing wave CAN be in the recording (if good omni condenser mikes or ribbons were used) but is ALWAYS present in the listening room too. External noise sources are usually not that big of a deal. Tracker organs almost(?) never produce deep fundamentals with real power, probably because of the low-pressure pipes. So the modern electric-action organs with the higher pressure pipes perform that service. An earthquake's initial acoustic waves traveling through the earth are pretty much the same as the big organ acoustic waves, although at a much lower frequency. It's not necessary to have walls around those earthquake waves to feel them, but some some reason it's necessary to have a large room with strong walls and no large openings to provide the reinforcement to be able to sustain and not lose the low frequencies in high fidelity. I'm not speaking of standing waves, which I think are the room resonance frequencies - I'm speaking of those waves that fit within the largest room boundary. So with a maximum room dimension of 17 feet I think you would be resonating at about 32.7 hz, or low C on the organ pedal. It would be better to have a longer room to avoid that resonance and get a better sound. But you still have to deal with the resonances at the other two dimensions.
|
|
rowuk
Been here a while!
Pain in the ass, ex-patriot yank living in the land of sauerkraut
Posts: 1,011
|
Post by rowuk on May 8, 2012 19:56:37 GMT
Higher pressure pipes? I don't think so. Pressure has NOTHING to do with the ability to reproduce low notes. The lowest frequency pipes are Gedackt and are not suitable for high pressure. The high pressure pipes like spanish trumpets are never longer than the 16 foot register.
The power is not based on the pipe pressure, rather the resonance of the pipe. Pipe organs get loud by adding multiple registers.
I play trumpet professionally in Germany and have a lot of experience with the real thing.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on May 8, 2012 22:10:49 GMT
Sounds like Robin is blowing his own trumpet again ! ;D
Only kidding Robin.
|
|
|
Post by dalethorn on May 8, 2012 22:56:30 GMT
Higher pressure pipes? I don't think so. Pressure has NOTHING to do with the ability to reproduce low notes. The lowest frequency pipes are Gedackt and are not suitable for high pressure. The high pressure pipes like spanish trumpets are never longer than the 16 foot register. The power is not based on the pipe pressure, rather the resonance of the pipe. Pipe organs get loud by adding multiple registers. I play trumpet professionally in Germany and have a lot of experience with the real thing. I can't relate to the theory or science here, since I don't delve into that. But having 100 or so pipe organ CD's and a long history of attending organ concerts, I have yet to hear a tracker organ produce power in the deep pedals like the modern organs do. Perhaps there is some other factor.
|
|
|
Post by dalethorn on May 9, 2012 5:26:11 GMT
Seeing as how this topic is High Resolution Files, and that other topic is "locked" in spite of new posts, I thought I should point out a possible discrepancy in that other topic, which of course applies here too since this topic is also hirez files.
Two or more music files on your computer that have the same identical contents will sound the same, unless the difference has nothing to do with the contents of the files.
|
|