[ramicio]

 
[size=medium]

Umm I don't see how you can do a test with bit depth with mp3.  Makes no sense to me...  It's still all getting dithered down to 320 kbps.
​

[/size]

 

[grawk]

16 bits should give you 96dB of dynamic range.  There's no chance your recordings are using that.  If you're playing things loud enough to HEAR 96dB of dynamic range, then you'd be deaf VERY quickly.

 

[spagetka]

this is very funny
 
If somebody did not pass the blind test, the result is wrong.
If somebody pass the blind test, the result is wrong.
 
I took the samples from http://www.linnrecords.com/linn-downloads-testfiles.aspx
 
@ranicio: I did comparision between mp3 (320) vs 16bit vs 24 bit
 

 

[dfkt]

Did you downsample the 24/96 yourself, with proper tools? Or did you use a readily available "normal"-definition file?
 
Edit: I see, they provide different formats on that site. I wouldn't trust those - I would definitely make my own conversions from the 24/96 to be 100% sure I get a quality transcode.

 

[grawk]

They have a vested interest in making 24/96 files seem better.  

 

[dfkt]

Indeed, that's why I was asking.
 
I conducted my own tests on that matter, and with a properly downsampled file there is absolutely no audible difference at all to my ears.

 

[nick_charles]

 
Quote:

this is very funny
 
If somebody did not pass the blind test, the result is wrong.
If somebody pass the blind test, the result is wrong.
 
I took the samples from http://www.linnrecords.com/linn-downloads-testfiles.aspx
 
@ranicio: I did comparision between mp3 (320) vs 16bit vs 24 bit
 

 
The Linn samples are different in more than just codec/bit-depth, I've tested them in Audacity and CEP, for starters the mp3 file is not quite the same length as the other two, in CEP the min levels are lower on the mp3 file by 4 - 5 db as compared to the flac files. also in CEP if you do a FR analysis on the mp3 and the 24 bit flac and compare the levels up to 20K there is an average difference of 4.22db (do a 8K FFT on the mp3 and a 16K on the 24 bit which gives you the same frequency points up to 22K). casually they look the same but they are very different in the audible spectrum.
 
To make sensible comparisons you must take the reference file and convert yourself either downsampling with dither or encoding to mp3.

 
In fairness to Linn the mp3 file is a very good encode it maintains high frequencies very well right up to 20K when it falls off the proverbial cliff.
 
NB statistically you really want to do at least 10 trials, Foobar has a very good ABX plug-in.
 

 

[spagetka]

If somebody can send me the reference files, I can make a blind test.
 
I have found Linn as extremely good with lot of beautiful recordings. My DAC1 accept "only" 24/96 via usb... I am using JRiver (WASAPI Event) and I have to say - great! 
 
 
 
Quote:

 
Quote:

this is very funny
 
If somebody did not pass the blind test, the result is wrong.
If somebody pass the blind test, the result is wrong.
 
I took the samples from http://www.linnrecords.com/linn-downloads-testfiles.aspx
 
@ranicio: I did comparision between mp3 (320) vs 16bit vs 24 bit
 

 
The Linn samples are different in more than just codec/bit-depth, I've tested them in Audacity and CEP, for starters the mp3 file is not quite the same length as the other two, in CEP the min levels are lower on the mp3 file by 4 - 5 db as compared to the flac files. also in CEP if you do a FR analysis on the mp3 and the 24 bit flac and compare the levels up to 20K there is an average difference of 4.22db (do a 8K FFT on the mp3 and a 16K on the 24 bit which gives you the same frequency points up to 22K). casually they look the same but they are very different in the audible spectrum.
 
To make sensible comparisons you must take the reference file and convert yourself either downsampling with dither or encoding to mp3.

 
In fairness to Linn the mp3 file is a very good encode it maintains high frequencies very well right up to 20K when it falls off the proverbial cliff.
 
NB statistically you really want to do at least 10 trials, Foobar has a very good ABX plug-in.
 

 

[nick_charles]

 
Quote:

If somebody can send me the reference files, I can make a blind test.
 
I have found Linn as extremely good with lot of beautiful recordings. My DAC1 accept "only" 24/96 via usb... I am using JRiver (WASAPI Event) and I have to say - great! 
  

http://www.divshare.com/download/13844696-99c]DivShare File - recit24bit.wav
 

http://www.divshare.com/download/13844697-e47]DivShare File - recit16bit-converted-to-mp3-then-back-to.wav
 

http://www.divshare.com/download/13844698-a98]DivShare File - recit16bit.wav
 
 
Above are links to 24 bit file, 16 bit file (downsampled) and 320K mp3 file all are in wav containers. Load any two into the FooBar ABX comparator and run 10 trials or more.

 

[monterto]

Fantastic write up!
 
I have a question that I'm a bit hung up about. This quote,
 
"In essence the theorem shows that an analog signal that has been sampled can be perfectly reconstructed from the samples."
 
Assuming I am not mis-understanding this then there are no advantages to listenning to vinyl vs CD?

 

[khaos974]

Could you post them on mediafire? I can't access them from China.
 
Quote:

Originally Posted by nick_charles /img/forum/go_quote.gif
 
http://www.divshare.com/download/13844696-99c]DivShare File - recit24bit.wav
http://www.divshare.com/download/13844697-e47]DivShare File - recit16bit-converted-to-mp3-then-back-to.wav
http://www.divshare.com/download/13844698-a98]DivShare File - recit16bit.wav
 
Above are links to 24 bit file, 16 bit file (downsampled) and 320K mp3 file all are in wav containers. Load any two into the FooBar ABX comparator and run 10 trials or more.

 
 
If your goal is fidelity to the master, there is no advantage to listening to vinyl instead of CDs.
Baring that, the questions of catalog content, different masterings, cost, emotional attachment, rituals... But if the question is purely about the accuracy to the recorded material, CDs beat vinyl by a fairly large margin.
Yes, the content can be theoretically perfectly reconstructed fro the samples up to 22.5 kHz.
 
Quote:

Originally Posted by monterto /img/forum/go_quote.gif
 
"In essence the theorem shows that an analog signal that has been sampled can be perfectly reconstructed from the samples."
 
Assuming I am not mis-understanding this then there are no advantages to listenning to vinyl vs CD?

.

 

[ramicio]

I just love how people think because something is supposedly proven mathematically, on paper, that it means it will work in real life.  Good luck with your theorems applying to physical electrons.

 

[khaos974]

I love how you interpret my words, I never said that reconstruction filters were perfect, did I?

 

[spagetka]

Thank you very much. The files were downloaded. I'll do the test as soon as possible.
 
 
BTW:

1. How can I verify that the files are not the same (only the name has changed)?
2. Is there ABX plug-in available for Jriver Media center?

 
 
Thank you.
 
 
Quote:

 
Quote:

If somebody can send me the reference files, I can make a blind test.
 
I have found Linn as extremely good with lot of beautiful recordings. My DAC1 accept "only" 24/96 via usb... I am using JRiver (WASAPI Event) and I have to say - great! 
  

http://www.divshare.com/download/13844696-99c]DivShare File - recit24bit.wav
 

http://www.divshare.com/download/13844697-e47]DivShare File - recit16bit-converted-to-mp3-then-back-to.wav
 

http://www.divshare.com/download/13844698-a98]DivShare File - recit16bit.wav
 
 
Above are links to 24 bit file, 16 bit file (downsampled) and 320K mp3 file all are in wav containers. Load any two into the FooBar ABX comparator and run 10 trials or more.

 

[monterto]

Quote:

If your goal is fidelity to the master, there is no advantage to listening to vinyl instead of CDs. Baring that, the questions of catalog content, different masterings, cost, emotional attachment, rituals... But if the question is purely about the accuracy to the recorded material, CDs beat vinyl by a fairly large margin.
Yes, the content can be theoretically perfectly reconstructed fro the samples up to 22.5 kHz.
.

This is very interesting, thank you for the insight. I suppose we all need to be more careful of the assumptions we make.