Difference between Analogue and Digital made visible. Not

[Netlist]

I have posted this at the Adobe forum but it would be difficult for many to follow as you can’t access without login.
So here goes:

Using Adobe Audition (or for those among us that still have Cooledit, I believe this will work just fine) I recorded a piece of LP music to HDD.
Then, I burned it on CD and ripped it again with the same program.
Purpose was of course to find out if any difference was measurable or could be made visible.
Audition has a nifty feature that can detract waveforms.
No need to say that both tracks have to be identical or close to or the thing gives un-interpretable results.
The original recording can be found here:
http://www.pixentral.com/show.php?pi...YbJrEbi6ufZfDx

The CD rip is here:
http://www.pixentral.com/show.php?pi...Te0rQwKVqJebA1

They look very identical and here are the measured differences:
Original recording:
Left Right
Min Sample Value: -12826 -21061
Max Sample Value: 12713 21244
Peak Amplitude: -8.14 dB -3.76 dB
Possibly Clipped: 0 0
DC Offset: -.015 -.015
Minimum RMS Power: -59.23 dB -58.95 dB
Maximum RMS Power: -15.55 dB -12.27 dB
Average RMS Power: -31.12 dB -24.68 dB
Total RMS Power: -29.47 dB -23.02 dB
Actual Bit Depth: 16 Bits 16 Bits

Using RMS Window of 50 ms

Rip:
Left Right
Min Sample Value: -12826 -21061
Max Sample Value: 12713 21244
Peak Amplitude: -8.14 dB -3.76 dB
Possibly Clipped: 0 0
DC Offset: -.015 -.015
Minimum RMS Power: -59.23 dB -58.95 dB
Maximum RMS Power: -15.55 dB -12.27 dB
Average RMS Power: -31.12 dB -24.68 dB
Total RMS Power: -29.47 dB -23.02 dB
Actual Bit Depth: 16 Bits 16 Bits

Using RMS Window of 50 ms

NIL, nothing, nada, nul difference. Not in time, not in amount of samples, not in phase.

So, both tracks should cancel out to zero or very close when deducted.
Here is the deducted waveform:
http://www.pixentral.com/show.php?pi...HP5C8m7lShSQ1D
Note the first 20 seconds of audio. They are indeed about 6db quieter. The rest seems to be simply added.

No, there is nothing wrong with the program or the settings as this is both waveforms added:
http://www.pixentral.com/show.php?pi...VjbVDVrBzhuHyf
Note again the first 20 seconds who appear to be about 6db louder.

If you are still reading, I’m puzzled by these results and unable to find an answer.
So are the guys at Adobe (for the time being)
More to come.

/Hugo

[darkfenriz]

You mean, that you ripped CD with CDROM or using CD player and sound card input?

[Netlist]

It's all PC work.
CD was ripped with CDRom.
I tried EAC and Audition to rip, both gave the same result.

/Hugo

[BlackCatSound]

You're missing some critical numbers. Such as sample count.

They may be identical in terms of peak values, offsets etc.. but if the CD ripper added a few extra zero samples at the start (and they do) then they won't line up in time and subtracting the two won't give a zero result.

[Netlist]

Tom, you might be right but interestingly, I recorded another track twice (from LP) and matched both as close as it possibly gets in start and ending of the pieces.
Both have exactly the same samples and start and stop at exactly the same time. (as exact as I can do it)
Comparing both Amplitude Statistics gives quite a bit of difference as seen below, anyway more than the ones in the above post which are identical.
Amazingly, the subtract function works pretty good, I have a clear picture of the differences in waveform from both samples.

Sample1
Left Right
Min Sample Value: -16144 -21447
Max Sample Value: 15936 19965
Peak Amplitude: -6.15 dB -3.68 dB
Possibly Clipped: 0 0
DC Offset: -.015 -.015
Minimum RMS Power: -54.59 dB -54.24 dB
Maximum RMS Power: -13.98 dB -12.22 dB
Average RMS Power: -29.59 dB -27.16 dB
Total RMS Power: -28.09 dB -25.54 dB
Actual Bit Depth: 16 Bits 16 Bits

Using RMS Window of 50 ms

Sample2

Left Right
Min Sample Value: -15976 -21645
Max Sample Value: 16076 20063
Peak Amplitude: -6.18 dB -3.6 dB
Possibly Clipped: 0 0
DC Offset: -.015 -.015
Minimum RMS Power: -53.37 dB -54.87 dB
Maximum RMS Power: -13.99 dB -12.17 dB
Average RMS Power: -29.59 dB -27.15 dB
Total RMS Power: -28.09 dB -25.53 dB
Actual Bit Depth: 16 Bits 16 Bits

Using RMS Window of 50 ms

I believe but will check again that the original recording and the rip match even closer.
If you really mean the amount of samples in both tracks, they are identical: 18763080

/Hugo

[darkfenriz]

At first it looked like fault in timing, like slowish phase fluctuation, but if it is all PC work, then there is no justification for it.

[Netlist]

I found the problem.
Although both tracks had the same sample count they were slightly mismatched in time.
I was able to find that with the ‘Find next zero crossing’ tool.
There is a difference of 132 samples between the first zero cross in both tracks.
I then took an equal amount of samples from that point on (18678327) in both tracks and now they match perfectly.
When detracting, the result is absolute silence. (-591db or as low as the graph can show) So the conclusion would be that burning audio on CDRom does in no way alter the music.
I believe I can claim that apart from the different medium they will be played from (cd player or HDD), it is impossible to hear any difference.
So, part of this acid test has been solved. http://www.diyaudio.com/forums/showt...486#post989486
The other part – proof with data – that an LP and the copy on CD sound equal (or not) is AFAIK virtually impossible.

/Hugo

[pinkmouse]

What I expected, but useful to know for certain.

[MikeB]

Yes, of course, unless you have real read/write errors (that don't happen with non defect medias) the music will not be altered. All these are digital copies, only errors can change the data.

So, why do they sometimes sound different ? One would expect no difference... The answer is jitter, "cheap" players seem to not reclock the signal received from the laser pickup.
This means, playing back exactly the same data played through different chains/medias can sound different.

A more interesting test would be to burn the wave onto a cdr, play it back on a cd-player an re-record it through SP/DIF.
Chances are 99.9999% that the result is again identical. (if soundcard does not resample to other samplerate, most cards can only work with 48khz and resample everything else... )

Mike

[Geek]

Originally posted by MikeB
So, why do they sometimes sound different ?

I know when I bought my Yamaha "Natural Sound" CD player with one of those 1-Bit DAC thingies, it was like listening to my CD collection all anew again. Amazing difference!