[Avi]

Quote:

Quote:

You're choosing between a full range high bitrate master and a regular CD with noise above the audible spectrum. I think that pretty much shows that you don't need those super audible frequencies or high bitrates.

 
+5.

 
Seems to be

 

[analogsurviver]

Interesting. Since I never manipulate the recording in any way, it sounded to me as described. Funny thing is that both me and another friend selected  a as the best, although we were both going on "wait a second, that c isn't bad - let me hear it once more" - goes on to show just how much we have been forced to listen to CD to de facto accept it as standard. b really sounds like MP3 vs better digital to me - since I never heard an recording manipulated in this way, by removing everything above just above audio band and injecting white noise instead, I equated it with the closest thing I know and that is MP3.
 
If it was not for the musical content of the CDs I like, I would have banned CD altogether. When still working in CD retail, customers who turned regular customers always kept me answering the same thing : Do you also find older (analog) recordings as better sounding than the modern (digital ) ones - despite all being on the CDs. No prizes for guessing the answer.
 
Thank you for the test. 

 

[xnor]

Quote:

Interesting. Since I never manipulate the recording in any way, it sounded to me as described. Funny thing is that both me and another friend selected  a as the best, although we were both going on "wait a second, that c isn't bad - let me hear it once more" - goes on to show just how much we have been forced to listen to CD to de facto accept it as standard. b really sounds like MP3 vs better digital to me - since I never heard an recording manipulated in this way, by removing everything above just above audio band and injecting white noise instead, I equated it with the closest thing I know and that is MP3.

You described B as the worst and A as the most natural, yet A and B are identical except for the reduction of bit depth to 16 bits for A.
 
Information below ~21 kHz was preserved in both cases and no lossy compression was used.

 

[chewy4]

Quote:

Interesting. Since I never manipulate the recording in any way, it sounded to me as described. Funny thing is that both me and another friend selected  a as the best, although we were both going on "wait a second, that c isn't bad - let me hear it once more" - goes on to show just how much we have been forced to listen to CD to de facto accept it as standard. b really sounds like MP3 vs better digital to me - since I never heard an recording manipulated in this way, by removing everything above just above audio band and injecting white noise instead, I equated it with the closest thing I know and that is MP3.
 
If it was not for the musical content of the CDs I like, I would have banned CD altogether. When still working in CD retail, customers who turned regular customers always kept me answering the same thing : Do you also find older (analog) recordings as better sounding than the modern (digital ) ones - despite all being on the CDs. No prizes for guessing the answer.
 
Thank you for the test. 

Interesting how you come to the conclusion that CD's are bad after picking CD quality as the best in a blind test...

 

[Averruncus]

Quote:

Information below ~21 kHz was preserved in both cases and no lossy compression was used.

Wait, how is that possible? You rendered the file twice after applying a filter and then dither with downsampling (not even symmetrical). Inverting the phase and comparing C and B shows some freq lost well below 20kHz (although theoretically inaudible). Isn't it possible that the whole process introduced errors or noise into the signal (once again, probably inaudible)?

 

[nick_charles]

Quote:

Wait, how is that possible? You rendered the file twice after applying a filter and then dither with downsampling (not even symmetrical). Inverting the phase and comparing C and B shows some freq lost well below 20kHz (although theoretically inaudible). Isn't it possible that the whole process introduced errors or noise into the signal (once again, probably inaudible)?

 
AFAICT The bottom line is that nobody was able to reliably, confidently and consistently identify the unmangled file from the two technically inferior ones, whatever errors may have been introduced only make this failure more interesting. That musical content above 22K can be replaced undetectably with random noise or that a (correct me if I am wrong) non-optimal dither procedure to lower the bitrate and thus SNR/dynamic range from 24 bits (theoretically 144db but realistically rather less) to 16 bits and (96db give or take) does not stick out like a sore thumb must give pause...

 

[stv014]

Quote:Originally Posted by Averruncus /img/forum/go_quote.gif

Inverting the phase and comparing C and B shows some freq lost well below 20kHz (although theoretically inaudible).

 
The difference below 20 kHz is at a very low level, and probably just resulted from 24-bit quantization, rounding errors, and maybe not perfectly flat frequency response near 20 kHz (even a level difference of only 0.0001 dB will result in a difference of only about 100 dB lower than the original signal, and that is without any phase errors).

 

[stv014]

Quote:

Spoiler: Warning%3A%20Spoiler!

Still, I did find B to be the worst, which makes me curious on how you filtered the freqs beyond 22KHz. When you resampled it to A, the resampler probably had a smoother filter Q and better aliasing compared to what you did in B, which I suspect made it sound better as the transients would sound more natural (bit depth probably made no difference). If not for B, I would probably not have been able to give a guess at all.Interesting test. I'm glad my work setup (fubar IV via USB -> FA-003) was able to give me enough fidelity for a guess.

 

Spoiler: Warning%3A%20Spoiler!

A was created from B by quantization to 16-bit using +/- 1 LSB triangular white noise dither. That adds a constant hiss to the sound. Other than that, there is no change.

 

[stv014]

Quote:

Interesting how you come to the conclusion that CD's are bad after picking CD quality as the best in a blind test...

 
Well, a.flac is actually still better than CD quality, because the 96 kHz sample rate made the dither noise quieter in the audio band, and while the ultrasonic content was degraded, it is not completely missing. For a more fair test, try my d.flac (posted here) in the foobar ABX comparator against c.flac; this was created by actually converting c.flac to a temporary 44.1/16 format file, and then converting that back to 96/24 (without using any information from the original c.flac), so it is guaranteed not to have any secret advantage over real CD audio.

 

[stv014]

Quote:

Do another one!  Once a week tests would be fun haha.

 
How about some loopback tests, using this sample, xnor's c.flac, or others if good ones are suggested ? Or various simulated DAC reconstruction filters (linear vs. minimum phase etc.) ? Or simulated quality degradation of various kinds (distortion, crosstalk, etc. at different levels) ?

 

[Todd R]

Quote:

 
AFAICT The bottom line is that nobody was able to reliably, confidently and consistently identify the unmangled file from the two technically inferior ones, whatever errors may have been introduced only make this failure more interesting. That musical content above 22K can be replaced undetectably with random noise or that a (correct me if I am wrong) non-optimal dither procedure to lower the bitrate and thus SNR/dynamic range from 24 bits (theoretically 144db but realistically rather less) to 16 bits and (96db give or take) does not stick out like a sore thumb must give pause...

 
Nick,
The question asked was for us to identify which file sounded best, not to pick out the un mangled file (which at the time we didn't know what if anything was done to those files).
 
I think a better way to run a test like this would be to present an un altered file as a control, so the testers know what the music sounded like originally.
Then the test would be to compare the control file to files A & B which might (or might not) have something about them changed and ask how they compare to the control file.
 
I did not know the music so I had no idea what it was supposed to sound like before it was altered, but I was able to pick the one that sounded best to me in spited of it technically being the "worst" of the group.
 
What I find interesting is you didn't hear any difference between the 3 files even though I thought B stood out as being quite different.

 

[bigshot]

Interesting how you come to the conclusion that CD's are bad after picking CD quality as the best in a blind test...

Some opinions are invested too heavily in to change. That would require thinking to process the data and coming to a new conclusion.

 

[bigshot]

I think a better way to run a test like this would be to present an un altered file as a control, so the testers know what the music sounded like originally.

The whole point of the test was, "Is there an audible difference between high bitrate and redbook?", not "Which one sounds most like X?" The choices were all over the map, which shows that not only are the two audibly identical, you can even toss in noise above the range of human hearing and no one will detect it. The test just had one beautiful red herring added in to throw cheaters off the track. That actually seemed to be completely successful. We all know that folks tried to cheat and it didn't help them. It even proved its own point... Inaudible frequencies are inaudible.

 

[nick_charles]

Quote:

 
Nick,
The question asked was for us to identify which file sounded best, not to pick out the un mangled file (which at the time we didn't know what if anything was done to those files).
 
I think a better way to run a test like this would be to present an un altered file as a control, so the testers know what the music sounded like originally.
Then the test would be to compare the control file to files A & B which might (or might not) have something about them changed and ask how they compare to the control file.
 
I did not know the music so I had no idea what it was supposed to sound like before it was altered, but I was able to pick the one that sounded best to me in spited of it technically being the "worst" of the group.
 
What I find interesting is you didn't hear any difference between the 3 files even though I thought B stood out as being quite different.

 
I would be more interested if you could correctly identify A/B as X/Y in a DBT regardless of what A or B are. You say you detected a difference but like so many others here you did this sighted as it were, you knew you were listening to A or B or C.
 
That you think B sounded different is far less interesting than if you could do this blind.
 
I could not detect a difference, and possibly for good reason - there is no difference that should rationally be humanly audible based on what we currently know - random noise above 22K - lowering of bitrate to16 bits neither of which to date has ever been shown empirically to be audible. To me accurate i.e blind detection of difference is a necessary precursor to questions of preference. If one cannot accurately detect that two things are different then a preference is nonsensical.

 

[analogsurviver]

Quote:

 
 
All 3 files have substantial content above 22khz - all 3 show the exact same and consistent pattern of gradual energy decline as frequency goes up, all three bottom off at about -107db at about 48khz consistent with 96Khz sampling but none show the falling off a cliff characteristic of MP3 or the brick wall decline to nothing at 22050 or thereabouts of 44.1 sampling... 

I thought this was a LISTENING test. I did not observe the samples on anything capable of displaying either frequency or dynamic range - something you obviously did.