|
|
|
urbo73 said:
... I for one hated CDs back in 1977! Because I knew they were coming. ...
Are we to assume you already knew what RBCDs were going to sound like when you decided not to like them in 1977?
|
|
|
Post by urbo73 January 7, 2010 (102 of 111)
|
|
Osbert Parsley said:
Are we to assume you already knew what RBCDs were going to sound like when you decided not to like them in 1977?
It was a joke of course, in response to how some people "hate" things for whatever reasons, sometimes ones that are completely irrational or nonsensical. My point, who cares, even if so? Someone may get angry at the color tie I'm wearing. No point discussing it!
|
|
|
Post by sordidman January 8, 2010 (103 of 111)
|
|
Wouldn't it be nice if it were that simple? Of course, none of this, has ever been about what people like and don't like. It has however, been all about one preaching religious zealot, unceasingly spamming the board with a barrage of unshared, narrow, hyperbolic extremism.
|
|
|
|
|
|
I just did some listening tests and measurements between 16/44 and 24/96 files, where the low-res file was made from the 24/96 original (violin concert, orginally recorded in DXD by 2L).
Using Foobar ABX comparator I first tried to compare the files, but the result was 11/20 and 12/20 which are no better than quessing.
Next I reversed the phase of one of the files and summed it with the other. The result was expected: nothing came out of the speakers/headphones. Not untill the volume was turned up about 20 dB louder that normal maximum level (peaks would have been about 130 dB SLP) a faint hf hiss could be heard.
I analyzed this difference signal, there was some scattered conversion artefacts here and there from 20 to 15000 Hz, but levels were below -120 dBFS, which means they are below the thresold of hearing even at loud listening levels. The 16/44 dithering hump centered at 18 kHz shows up quite clearly, but the average level was -70 dBFS, which combined to the sensitivity curve of hearing at that frequency means it is almost 90 dB below the peak values of music. If somebody really can hear it mixed with full signal, good for them, I can not. Even the overamplified difference hiss carried absolutelly no musical content in it, it was just a steady hiss. This is no surprice as it is just noise shaped dither, not the difference in musical content beween the files.
I you have hi-res files and a PC and some curiosity it is possible to do these tests with free tools, Foobar and Reaper (30 day free demo for the later).
For monitoring Crane Song Avocet monitor controller with 24/96 DAC, Adam S3A monitors, Hedd CanAmp and Sennheiser HD800 headphones were used. PC audio interface was Prism Orpheus.
|
|
|
Post by steviev January 10, 2010 (105 of 111)
|
|
jullepoika said:
Using Foobar ABX comparator I first tried to compare the files, but the result was 11/20 and 12/20 which are no better than quessing.
Thanks for sharing, and thanks for having the courage and integrity to do a blind comparison. I can only say "ditto" to your conclusion -- my accuracy rate was around 55% as well.
Viva multichannnel!
|
|
|
|
|
|
Doing an honest blind test saves one from so much grief. No more reason to try the reach the useless goal of ever higher resolution files for no real audible reason what so ever!
|
|
|
|
|
|
jullepoika said:
Next I reversed the phase of one of the files and summed it with the other. The result was expected: nothing came out of the speakers/headphones. Not untill the volume was turned up about 20 dB louder that normal maximum level (peaks would have been about 130 dB SLP) a faint hf hiss could be heard.
I analyzed this difference signal, there was some scattered conversion artefacts here and there from 20 to 15000 Hz, but levels were below -120 dBFS, which means they are below the thresold of hearing even at loud listening levels. The 16/44 dithering hump centered at 18 kHz shows up quite clearly, but the average level was -70 dBFS, which combined to the sensitivity curve of hearing at that frequency means it is almost 90 dB below the peak values of music. If somebody really can hear it mixed with full signal, good for them, I can not. Even the overamplified difference hiss carried absolutelly no musical content in it, it was just a steady hiss. This is no surprice as it is just noise shaped dither, not the difference in musical content beween the files.
I tend to agree with you that stereo hi rez PCM or DSD playback does not itself generate a significant audible difference over RBCD. This has been my own experience. However, I also believe ABX testing has its threshold below which the sorts of small audible differences that exist today in audio (speakers excepted, of course) are not reliably dicernable. So, it is not a surprise to me to that you were unable to identify the differences beyond chance.
I thought your inverted phase test was brilliant and does confim small differences. However, I am a bit troubled by the "dithering hump" at -70 db. I am no expert, but I would think that dithering involves only a few least significant bits, and would therefore yield an even smaller value than this.
Mightn't -70 db be audible to some people in some systems some of the time?
Might this "hump" not also include some residue of the infamous RBCD brick wall filter in the audible band, e.g, ripple, pre-ringing, time-domain irregularities? We know these exist in theory, so I would expect them to show up in your analytical data somehere. As I have stated elsewhere in this forum, my hypothesis has been that hi rez has two main advantages over RBCD: (1.)low level detail from the greater, up to 24-bit headroom and (2.)freedom from artifacts caused by the 22K filter, because filtering in hi rez is done so far above the audible band. So, in my view there is still the potential for small audible gains via hi rez.
I think it might also be true that some CD's recorded in the 44k/16 format, unlike your test, might provide a double dose of any filter artifacts: once on the A>D and again on the D>A. I am influenced here by Bob Stuart's thinking at Meridian and his new apodizing filter for RBCD that purportedly alleviates many audible consequences of these filter artifacts caused by both recording and playback.
|
|
|
Post by canonical January 12, 2010 (108 of 111)
|
|
jullepoika said:
Using Foobar ABX comparator I first tried to compare the files, but the result was 11/20 and 12/20 which are no better than quessing.
..... This is no surprice ...
... Sennheiser HD800 headphones were used. PC audio interface was Prism Orpheus.
Hi Petrus.
|
|
|
|
|
|
Dithering comes in many flavors. "Straight" dither might be even less than one bit deep and spread out evenly across all frequences, noise shaped dither which is now always used pushes this noise to frequences where the ear is less sensetive. An analogy: you have a dusty livingroom floor and your mother-in-law is coming to a surprice visit. Solution: sweep the dust to shady corners where she can not see it. The amount of dust remains the same, but is invisble.
This dither "hump" is the dustpile. -70dBFS might sound (figuratively) as relativelly high, but when mixed with strong signal and because it lies where the ear is not very sensetive it can not be heard (at least by me). Even if I listen to pure dither (inversion-summing test) this signal is so low even at normally loud listening levels that it hides itself to the ambient noise in the listening room. Actually this level is about the same as the low dead-quiet level of an extremely high dynamic range recording, but at a frequency which is difficult to hear.
With modern AD and DA coverters the filtering problems lay mostly with the recording. There is really no way to avoid the antialiasing filter problem, but one way to minimize it is to record at higher sample rates and use software sample rate conversion later in the chain. There it is at least possible to test different configurations in stead of relying on the fixed filters of the recoder. As practically all players now upsample before DA and lowpass filtering there is no filter problems there anymore. Filtering is done so far above the human hearing range that the filter riples do not reach the audible frequencies.
|
|
|
|
|
|
canonical said:
Hi Petrus.
Hi there! Nice to be back.
Zeus got me already several weeks ago, you are not reading the threads all that sharply...
|
|
