Thread: Alert: All Downloaders of Hi Res files.

Posts: 76
Page: prev 1 2 3 4 5 6 7 8 next

Post by audioholik May 11, 2011 (21 of 76)
bissie said:

This, I think, is most unfair. We had nothing to do with this whatsoever. We didn't even deliver the material to HDTracks, and we certainly were never asked or consulted. I have said this often enough, so I do think it is incumbent on you to phrase your posts less unfairly.

Robert

Let me clarify then - the only responsible for selling BIS recordings(recorded at 44kHz) as 88kHz was HDTracks.com, that they continue to sell any upsampled files is inexcusable IMO. That was/is my only point.

Post by bissie May 11, 2011 (22 of 76)
audioholik said:

Let me clarify then - the only responsible for selling BIS recordings(recorded at 44kHz) as 88kHz was HDTracks.com, that they continue to sell any upsampled files is inexcusable IMO. That was/is my only point.

Accepted. This matter has beed debated so long, and in such vile language that I am (hopefully understandably) a tad sensitive about it.

BTW we now seem, after some investments and soul-searching, to be in the position to record everything - except PERHAPS the very largest recordings - in true hi-res. Possibly this theme can be laid at rest now for good.

Robert

Post by bissie May 11, 2011 (23 of 76)
Arnaldo said:

While this is very good news indeed, it would be even better with specific recording info for each release. As consumers, we have no way of knowing when and how a particular recording was actually made, or if it falls under the "the very largest recordings" category.

Either way, I truly appreciate the effort, both in terms of the sizable financial investment as well as to the very touching "soul-searching" comment...

Moved to the BS Thread
(edit) No Freudian slip here, I actually meant BIS thread :-)

Robert

Post by canonical May 11, 2011 (24 of 76)
The issues raised do not only relate to download sites ... the same issues exist for some SACDs. But downloads do seem to be more prone to these problems, perhaps because they are often sourced from physical SACD discs, and so they not only have any problems that might exist on the SACD, but also additional problems created by downsampling SACDs to download files.

I haven't read the article ... only the itrax link comment on it referred above ... but would quickly make two comments:

1. The author of the itrax comment piece might mean well, and I appreciate and agree with his intentions (especially that all SACDs and download files should be properly labelled), but he makes comments such as:

. . . . . . . . . . . "Anything that goes back to the days of analog tape shouldn't be "upsampled" and sold for a premium price"

... that suggest he is frankly clueless on technical matters. Analog tape to SACD is not 'upsampling' --- SACD is still lossy digital ... just a lot less lossy than Analog to CD.


2. Spectrum analysis showing cut-offs at 22kHz does not necessarily mean that a recording was made at 44.1kHz. It might have been made at say 96 kHz ... and then a low-pass filter imposed that cuts off above say 22kHz. As I understand it, this would still leave a high detailed sampling rate for the information below 22 kHz, just truncating the signal at 22 kHz ... and I believe this is quite common in the recording industry (though I am not sure why). Perhaps someone familiar with the practise could comment on why it is used...

Post by AmonRa May 11, 2011 (25 of 76)
canonical said:

1. The author of the itrax comment piece might mean well, and I appreciate and agree with his intentions (especially that all SACDs and download files should be properly labelled), but he makes comments such as:

. . . . . . . . . . . "Anything that goes back to the days of analog tape shouldn't be "upsampled" and sold for a premium price"

... that suggest he is frankly clueless on technical matters. Analog tape to SACD is not 'upsampling' --- SACD is still lossy digital ... just a lot less lossy than Analog to CD.


2. Spectrum analysis showing cut-offs at 22kHz does not necessarily mean that a recording was made at 44.1kHz. It might have been made at say 96 kHz ... and then a low-pass filter imposed that cuts off above say 22kHz. As I understand it, this would still leave a high detailed sampling rate for the information below 22 kHz, just truncating the signal at 22 kHz ... and I believe this is quite common in the recording industry (though I am not sure why). Perhaps someone familiar with the practise could comment on why it is used...

1) Analog tape is much, much more "lossy" than even CD. Everything* in an analog tape can be preserved on a CD, no need to use SACD.

2) A "natural" high resolution recording with extended high frequency range past 20 kHz shows a gently downsloping frequency content** form about 3000 Hz up. There are no sudden cutoffs. If the frequency analysis shows a cutoff at 22 khz the only explanation is an unreversable conversion to 44.1 kHz PCM at some point. There are no tecnical reasons to filter upper frequencies in high resolution recordings. Natural levels above 20 kHz are so low that they do not eat up amplifier power or burn supertweeters. If you think filtering at 22 kHz somehow leaves high frequency content hiding somewhere below 22 kHz you are understanding wrong. Filtering FH does just that: filters it away, sands the sharp angles in the waveforms round, destroys detail. BUT: we do not hear this detail, so we do not miss it, no harm done.***

*) In theory there could be some content above 22 kHz on a MODERN analog tape recording recorded with MODERN microphones, but not in real life. Tapes from the fifties-eighties: forget it.

**) DSD exhibits a rising frequency content above 25-30 kHz, but this is not real musical content (nature does not work this way), it is DSD noise shaping. This should be filtered out before amplification, it is useless and even harmfull to equipment.

***) simplification, this. It is not the HF that we miss and loose quality, it might be the sharp lowpass filters which might alter the sound. For this reason all decent players upsample the signal before filtering.

A natural sound is truly lossless. Analog tape is a crude approximation, an analogy, of the real thing, far from lossless (I recently posted the specs of the best professional tape recorder available: it is downright lousy compared to a decent CD quality recorder costing 90% less). Sometimes I feel that people mix up these things thinking that an analog recording is somehow truer to the nature. It is not, the only criteron are how well the real thing is preserved, no matter what the method is. "CD quality" vastly surpasses analog tape and direct cut LP in every possible way (exept quater of an octave in HF).

Post by Jay-dub May 11, 2011 (26 of 76)
AmonRa said:

1) Analog tape is much, much more "lossy" than even CD. Everything* in an analog tape can be preserved on a CD, no need to use SACD.


*) In theory there could be some content above 22 kHz on a MODERN analog tape recording recorded with MODERN microphones, but not in real life. Tapes from the fifties-eighties: forget it.

I frequently look at spectrograms of digital music files in order to get an estimate of the effective bandwidth of the recording (defined as the frequency above which musical features are no longer visible, being buried under the noise power spectrum even during louder passages). For CD's mastered with psychoacoustic (inverse A-weighted) noise-shaping, the effective bandwidth is about 17kHz. For HDTracks downloads made from analogue recordings of the '70's, it's typically about 28 kHz.

Recordings made at 44.1kHz can sound very good if the antialiasing filter used was good (i.e. phase-linear, with a cutoff frequency between 20 and 21.5 kHz). Very few CD's fit this description. Those made in the '80's all used analogue anti-aliasing filters which caused phase distortion. More recent ones usually were recorded with digital decimating filters that are down 3dB at 22.05kHz and leave very significant aliasing in most cases down to 21-21.5 kHz, but in some cases 20.5 kHz. I keep on copying CD's into my computer and applying lowpass filters to remove these defects from the sound files, with the result of greatly increased listening pleasure.

Post by AmonRa May 11, 2011 (27 of 76)
Jay-dub said:

I frequently look at spectrograms of digital music files in order to get an estimate of the effective bandwidth of the recording (defined as the frequency above which musical features are no longer visible, being buried under the noise power spectrum even during louder passages). For CD's mastered with psychoacoustic (inverse A-weighted) noise-shaping, the effective bandwidth is about 17kHz. For HDTracks downloads made from analogue recordings of the '70's, it's typically about 28 kHz.

Recordings made at 44.1kHz can sound very good if the antialiasing filter used was good (i.e. phase-linear, with a cutoff frequency between 20 and 21.5 kHz). Very few CD's fit this description. Those made in the '80's all used analogue anti-aliasing filters which caused phase distortion. More recent ones usually were recorded with digital decimating filters that are down 3dB at 22.05kHz and leave very significant aliasing in most cases down to 21-21.5 kHz, but in some cases 20.5 kHz. I keep on copying CD's into my computer and applying lowpass filters to remove these defects from the sound files, with the result of greatly increased listening pleasure.

The acid test is to compare a loud passage analysis to a ambient noise passage and see it the HF content stays the same. Only if not, there is some musical signal related stuff up there. Even then it might not be upper harmonics, but rather harmonic distortions and tape noise. Old microphones did not put out much above 20 kHz, they typically start to cut from 16 kHz, so I am skeptical.

New software based sample rate converters are much, much better than old ones, and what is best, they are either free like SOX utility, or cost only about $50 like iZotope included in Sample Manager or Wave Editor. They even equal or beat hardware units costing thousands, like Weiss Saragon. New Pyramix has apodizing filters for SRC, I have not heard it yet or read any reports, it could be even better. So the problems with phase altering filters are not nearly as bad as it used to be at the infancy of CD. Every engineer worth his salt uses these tools now for mastering CDs.

Post by Disbeliever May 11, 2011 (28 of 76)
Jay-dub said:


Recordings made at 44.1kHz can sound very good if the antialiasing filter used was good (i.e. phase-linear, with a cutoff frequency between 20 and 21.5 kHz). Very few CD's fit this description. Those made in the '80's all used analogue anti-aliasing filters which caused phase distortion. More recent ones usually were recorded with digital decimating filters that are down 3dB at 22.05kHz and leave very significant aliasing in most cases down to 21-21.5 kHz, but in some cases 20.5 kHz. I keep on copying CD's into my computer and applying lowpass filters to remove these defects from the sound files, with the result of greatly increased listening pleasure.

According to one of the designers of the Sony DA5400ES, 'a large number of CD collectors have discs they've discarded because the sound quality is very poor, particularly early releases. Those early analogue-digital converters wern't very accurate. They fractured the signal, causing damage. With the DA5400ES AVR, we can compensate for the early brick-wall filter used in first-gen CD mastering; its audio processing gives a more forgiving roll-off and removes obvious faults created by the D-A process'

Post by canonical May 12, 2011 (29 of 76)
Shame - AmonRa (Petrus w/e) still doesn't understand that a digital sample is just that: a sample of the analog original. Tut/ Join the dots, AmonRa . ° . ° . . .

Post by AmonRa May 12, 2011 (30 of 76)
canonical said:

Shame - AmonRa (Petrus w/e) still doesn't understand that a digital sample is just that: a sample of the analog original. Tut/ Joins the dots, AmonRa . ° . ° . . .

The measure of the quality of the reproduction is how well the output waveform follows that of the input signal. The fact is that this digital method of taking samples fast enough and then "joining the dots" (if you wish) and properly low pass filtering the output produces a vastly better approximation of the input signal than any analog recording method. The signal which comes out of the digital-to-analog converter is just as continuous as the original. If it was not, and better than what analog tape recorders can produce, we would still be using tape. Guess what? Nobody even makes studio tape recorders anymore except apparently Otari and only one model!

It would be very easy to make a loopback test, maybe 20 iterations, by re-recording an analog tape and digital recording, and compare the results. My educated guess would be that the 20:th generation digital recording (CD quality) would sound quite good, even very good, and the analog counterpart would be a cloudy, hissy mess. What would your uneducated guess be, canonical?

Page: prev 1 2 3 4 5 6 7 8 next

Closed