And most of the folks who do it for a living (recording and mastering engineers, producers) seem to agree that Red Book is a step down in quality from the master. Many also think that DSD and some high bitrate PCM gets very close to the quality of the master.
When you do it day in, day out year after year, you tend to get rather good at spotting those things. Scientific proof? No. But an important enough trend to wonder why.
When you do it day in, day out year after year, you tend to get rather good at spotting those things. Scientific proof? No. But an important enough trend to wonder why.
labjr: Standards and back-compatibility. That's not forever, but it sure drives things today. And it will take some evidence that there's an audible improvement if higher sample rates are to penetrate markets beyond a tiny niche.
@P10: IOW, no, there is no test that anyone has published that shows audibility.
@P10: IOW, no, there is no test that anyone has published that shows audibility.
And since you can't prove a negative, there never will be. You're safe in that belief.
There have been quite a few null tests published, some in refereed journals, and quite a few more that have not been (like my own). You're familiar with many of them, I'm sure, but I understand that you dismiss them all. That's your right, keep on hoping that sometime, somewhere, somehow, someone can find an audible difference.
I wouldn't bet on it, though. 20 years of CD and it hasn't happened, despite the large emotional baggage carried by a tiny minority.
Our audio club has done numerous comparisons using great gear, etc., and so far really good vinyl has trumped everything except "The Tape Project" master tapes.
One of our members owns a SOTA recording/mastering studio and as I've mentioned before, burned two discs off of his files, one using a Plextor CD burner and another disc using an LG Blu-Ray burner. He was amazed that the Blu-Ray copy was so much better sounding that he went back and compared the discs and could find no differences! It was only when he examined them under an electron microscope at the UW Hospital that he noticed that the edges of the pits on the CD burner's copy were ragged and "frayed" looking, while the pits on the Blu-Ray burned disc were sharp, clean and precise. All the ones and zeros were there, but...........?
Best Regards,
TerryO
It actually is a spiral with several levels of modulation on Red Book (maybe 11?) and from that it translates into digital words. Its not like a hard disk. I think I remember that from an old HFNRR article. Correct me if I remember cuckoo.
Cite?
The Lipshitz tests suffered from level issues- Lipshitz heard the difference, but Tiefenbrun didn't. And Lipshitz remarked that he only heard the difference when there was no music playing.
I was referring to a well-known test where a much better than null result was witnessed but then dismissed. The reports of some of these tests don't always include all of the raw data so it is difficult to analyze the interpretation made by persons who write the papers. It's true that I don't know whether aberrations were set aside or not in the tests you mentioned but I would like to know if anyone has seen the raw data from them.
John
Two questions:
1) what phase anomalies? Post-1990 ADCs have linear phase AA filters.
2) how can you claim the above without reference to unambiguous proof that a bandwidth of 20kHz is not sufficient, all else being executed as it ought to be?
1/ What is a linear phase AA filter? Anything analog to keep 22+k out of the ADC is going to have phase anomalies way down into the midrange
2/ we were talking about a CD being able to exactly duplicate a master tape. A master tape will have response > 20k.
Further Kuncher (in 2 different experiments and his results independently duplicated) has shown that time response on the order of 5 usec is the current measured detection threshold in humans. It takes more than 20k to get there.
dave
Delta-sigma ADCs, as used since 1990, have digital AA filters. These are all linear phase. If there is something wrong with 44.1kHz sampling it is not going to be related to phase response.
Is it fair to assume that we were talking about the sound, or were we just measuring?
If the former, then we need a proof that audio above 20kHz matters to us humanoids.
Now, up to ten years ago I firmly, just like you and perhaps most of the people here, believed that an ultrasonic system response would be very nice and desirable.
But then I started to investigate and found that there really wasn't any unambiguous proof for the audibility of ultrasonics, whether through direct detection or through spooky interaction with the audible range.
There were of course (many) claims by often knowledgeable people that more sounded better than less. Referring now mainly to mastering engineers, people generally closer to the source than use mere consumers. These claims, of course, carry some validity. But closer scrutiny then revealed that the majority of DAW software, especially back then, employed decidely sub-par sample rate conversion software and tended to ignore the more subtle repercussions of working with sampled data.
If A sounds better than B, and A has a higher bandwidth, then this is insufficient proof that bandwidth on its own does the trick. Not when both bandwidths exceed the limits of human hearing and when not all other potential differences between A and B have been investigated.
In the academic worlds, these of auditory percetion and of audio (AES), there have been a number of proofs, all IMO, pretty weak except when bone conduction was involved, but really, bone conducted sound is utterly irrelevant for the replay of music in a living room. Luckily so.
The recent work by Amandine Pras (just a blind listening test of recordings at several sample rates) yielded a positive, but this is weak again since it seems that the several recordings did not follow the exact same signal path, and, where an SRC was involved, it was said to be the one in the Pyramix DAW, an SRC which is known to be of suspect performance.
On the other hand we have experiments by the likes of BAS and Lipshitz that indicate the transparency of a 44.1kHz link.
Guess what: I reject these too. Lipshitz is an excellent academic and has done a lot of good for the audio community, but that single test with Tiefenbrun seemed in a setup that was less than ideal, and with too few (younger) subjects anyway.
As for the BAS and the likes of the people that crowd Hydrogen (like Arnogant Krueger, who displays there such an ignorance of signal theory that it becomes funny!), they have a pre-set agenda which is completely at odds with those who seek the truth, whatever that may turn out to be.
So, in short, I'd welcome a conclusive proof of the relevance of ultrasonics, but I'm not holding my breath anymore and I prefer now to concentrate on CD-done-right (and LP-done-right, just received my new Lyra!)
1) Have Kunchur's results been duplicated by a third party? Really? Please show us the evidence then.
2) It is trivial to pass 5us features through a 44.1kHz-sampled channel. Kunchur's experimental results may have some relevance for auditory perception, as they seem to refine the knowledge about some specific thresholds of hearing. But where he errs massively is in his conclusion with respect to sampling. He does so twice: first in the actual experiments, which would have yielded the same results if he passed the signals through a properly-done 44.1kHz channel (he used 7kHz as a fundamental, it should have been 8kHz to obtain something conclusive, do the math yourself), and then in claiming that 44.1kHz could not pass 5us features, which is blatantly wrong and soooo easy to verify by anyone with DAW software.
You actually get phase anomalies as you approach 20.05kHz, and at the limit of 20.05kHz you have no phase relation between the source and the digital source at all.
This is inevitable because at this frequency you have 2 points per waveform cycle. These two points are synthesised to the march of the master digital clock, a clock that is independent of the music.
Additionally the amplitude of this waveform is fixed as roughly the peak difference between these two points: the DAC does not know any different - it cannot, so amplitude information is also lost.
For these reasons perhaps it's best to consider the CD as a lossy format, the loss of information occurs beyond most people's hearing: but it's still lost. IMO a decent retimer/upsampler (and declipper these days) is the best way to decode a CD to get the most out of it: but there is still missing information and inevitable undersampling at the top.
An externally hosted image should be here but it was not working when we last tested it.
So lets not push the full 20(.05)kHz bandwidth: by definition it is less.
My personal despair with record companies is that the 24/192 rates have not been pushed: Intel and AMD made a fortune out of pushing clock speeds - even when it was patently clear that the P4's high clock speed wasn't cutting it to get the work done. The benefit to the industry and record companies is clear: you end up with bigger numbers to push (a marketeers dream) and you also end up with huge file sizes: far more laborious to download and store, far easier to buy on DVD. Hell, they could even sell 3 copies of the same track - the 16/44, 24/96 and 24/192.
I guess they prefer bleating about home taping killing music..
In fact quite why we are still playing obsolete silver discs in 2010 when computers abound I have no idea.
I know why we are still playing LPs though 😉
Hi,
do you mean that an oscillation in an amplifier at 30 Khz,or a 10 db resonance,deliberately, introduced at 25 Khz ,is audible,even though it is well out of the accepted human hearing range?
B.L
- Status
- Not open for further replies.