I read this morning, not for the first time, that some people believe that when you listen to a classical recording that was transferred from a 78, the resultant surface noise gives the impression that you are getting more high frequency information. Is there any way to prove this one way or another?
(http://ecx.images-amazon.com/images/I/41PXK2DDW5L._SL500_AA300_.jpg)(http://ecx.images-amazon.com/images/I/51y2IXwwIhL._SL500_AA300_.jpg)(http://ecx.images-amazon.com/images/I/2192W54N7TL._SL500_AA300_.jpg)
I have uploaded WAV samples of three CD transfers that were made of the historic Artur Schnabel's Beethoven sonata performances. The samples include one of EMI, Naxos and Pearl (OOP.) I think they clearly show the differences in mastering approaches that each of these labels took for these performances.
If you wish, you can download the samples here. (http://www.mediafire.com/?sharekey=45732805cf458166c74064dd4c81b78ae18e115634b0aa6f)
If you guys get a chance, could you please download these three samples? They are only 45 seconds long each and I love to know your impressions. There is a debate that has been going on about the SQ of these elsewhere on the internet and I would like to get to the bottom of this. I will tell you that the Naxos and Pearl were transferred from 78s and the EMI was transferred from a vinyl (much quieter) 78 that was reportedly made directly from the metal master. So the extra high frequency one hears on the Naxos or the Pearl some think is just an illusion.
What do you think?
To me the EMI (the one I have) and Naxos were pretty close, possibly the EMI slightly less mid-rangey, whereas the Pearl had so much more noise it would not be a recording I would want to use for these performances.
Quote from: George on June 13, 2010, 01:10:21 PM
I read this morning, not for the first time, that some people believe that when you listen to a classical recording that was transferred from a 78, the resultant surface noise gives the impression that you are getting more high frequency information. Is there any way to prove this one way or another?
There certainly is. One can look at the frequency spectra of the recordings, from which it would be relatively straightforward to judge the frequency at which surface noise is burying high frequency information. Unfortunately, this won't convince anyone, since claims made by "audiophiles" are not generally constrained by petty things like the laws of physics or mathematics. 8)
Quote from: Franco on June 13, 2010, 01:50:29 PM
To me the EMI (the one I have) and Naxos were pretty close, possibly the EMI slightly less mid-rangey, whereas the Pearl had so much more noise it would not be a recording I would want to use for these performances.
Thanks for listening and posting, Franco! :)
So you don't prefer the Naxos to the EMI?
I agree that the Pearl is simply too noisy. They gave me headaches, so I sold them off to a good friend who wasn't similarly bothered by them.
Quote from: George on June 13, 2010, 01:58:46 PM
So you don't prefer the Naxos to the EMI?
I could not tell much difference, granted I did not listen to them very loudly or over headphones, so I may not have be able to discover the subtle differences.
Quote from: Scarpia on June 13, 2010, 01:55:32 PM
There certainly is. One can look at the frequency spectra of the recordings, from which it would be relatively straightforward to judge the frequency at which surface noise is burying high frequency information.
OK, what can you tell from these images? For larger image,
click here. (http://img149.imageshack.us/img149/6533/schnabelspectrum.png)
(http://img149.imageshack.us/img149/6533/schnabelspectrum.png)
Quote from: Franco on June 13, 2010, 02:36:59 PM
I could not tell much difference, granted I did not listen to them very loudly or over headphones, so I may not have be able to discover the subtle differences.
Ok, thanks. To me, the difference between them is not subtle.
Looking at the wave files, it looks like Naxos did not filter out as much of the mid frequencies allowing more noise but also supply some more mid-high audio above what is in the EMI. However, in my original comment I said the EMI was less mid-rangey, something that I found preferable to the Naxos.
I've read that people think the Naxos is better - but not to me.
Quote from: Franco on June 13, 2010, 03:00:42 PM
Looking at the wave files, it looks like Naxos did not filter out as much of the mid frequencies allowing more noise but also supply some more mid-high audio above what is in the EMI. However, in my original comment I said the EMI was less mid-rangey, something that I found preferable to the Naxos.
Thanks!
Quote from: George on June 13, 2010, 01:58:46 PM
Thanks for listening and posting, Franco! :)
So you don't prefer the Naxos to the EMI?
I agree that the Pearl is simply too noisy. They gave me headaches, so I sold them off to a good friend who wasn't similarly bothered by them.
Good friend checking in! :)
See PM. :)
Quote from: George on June 13, 2010, 02:44:02 PM
OK, what can you tell from these images? For larger image, click here. (http://img149.imageshack.us/img149/6533/schnabelspectrum.png)
(http://img149.imageshack.us/img149/6533/schnabelspectrum.png)
I would do a different sort of analysis (I have downloaded the wave files and may do it if I find the time). However, it is clear from your posted graphs that the Pearl recording has a lot of continuous noise in the 5-10 kHz band. If we assume that the 5-10 kHz response of the piano was being recorded without loss (in itself doubtful) we can estimate how much sound would have been recorded in that 5-10 kHz band by extrapolating from the low frequency output of the piano. I think the conclusion of such an analysis would be that the sound produced by the piano in that band would be
far weaker than the continuous noise. However, the ear is not a microphone. It is possible that the ear can distinguish the high frequency overtones of the piano because they are in phase with the audible low frequency tones.
It would be fun to test that hypothesis. I think I will go to the pearl file, and prepare a version in which the phase of the high frequency portion has been scrambled. If there is no information there, it should make no difference. If there is, the scrambled version should sound worse.