Can you hear a difference between the top analog preamp and a straight wire? Listening test

[pkane]

DW has bugs.

If you would point them out, I could fix them. So far here, I don't see bugs, I see incorrect use of the software.

 

[pkane]

I get -60-something without matching (just "Show" button) and -80-something with matching ("Match" button).


These are mono files. Weird that there is a difference in correlated null and jitter shown in the status at the bottom.

I get the same values both for L and R but different than yours
View attachment 373086
(@pkane ?)

Incorrect settings, I suspect.

-80.5dB is the null if you use these (all linear) settings:

 

[Sokel]

Incorrect settings, I suspect.

-80.5dB is the null if you use these (all linear) settings:

View attachment 373109

For me it was the "measure simple waveforms" that messed it,it was left there after I did measured simple waveforms couple of days ago and prompted me to engage it.
I though it would reverse the setting after and I didn't check it

 

[pkane]

For me it was the "measure simple waveforms" that messed it,it was left there after I did measured simple waveforms couple of days ago and prompted me to engage it.
I though it would reverse the setting after and I didn't check it

Yes, that setting is not meant for music but for simple periodic test signals, like a single frequency sine wave.

 

[pma]

If you would point them out, I could fix them. So far here, I don't see bugs, I see incorrect use of the software.

I get unpredictable single sample drift, like shot noise, probably depending on processor load. This happens in this test with DW, but never with Adobe Audition direct subtraction. The drift of the single sample is more than 100 LSB levels high. Sounds like a click, which is non-existent in the test files.

 

[danadam]

Sounds like a click, which is non-existent in the test files.

There is a discontinuity in both files:

DW's subsample matching obviously won't like it:

 

[danadam]

Null of -80dB:
View attachment 373108

Mostly the same here with the exception of "correlated null":

No issues with phase or frequency response:

My frequency response looks slightly different:

 

[pma]

I am talking about file difference, and there is no discontinuity in the file created by subtraction. So I am asked to go to AA and show the amplified raw difference?? You guys better post your ABX protocol from 16 trials, that is the thread topic! Stop speculations, please.

 

[pkane]

I am talking about file difference, and there is no discontinuity in the file created by subtraction. So I am asked to go to AA and show the amplified raw difference?? You guys better post your ABX protocol from 16 trials, that is the thread topic! Stop speculations, please.

The discontinuity is in the original files, there's no speculation

 

[Blumlein 88]

Incorrect settings, I suspect.

-80.5dB is the null if you use these (all linear) settings:

View attachment 373109

Those are the settings I used and I see Paul, Sokel and myself had the same result out to 2 decimal places.

 

[Blumlein 88]

I get unpredictable single sample drift, like shot noise, probably depending on processor load. This happens in this test with DW, but never with Adobe Audition direct subtraction. The drift of the single sample is more than 100 LSB levels high. Sounds like a click, which is non-existent in the test files.

When I used Deltawave the difference file has a click when amplified at that 41.3 second point where the discontinuity is. When I do a straight subtraction in Audacity it is still there though more of sharp thump than a click. When I do this in Reaper it is a thump like in Audacity.

Now if I turn off "Correct clock drift" and "Subsample Offset" in Deltawave then I get the thump like in Audacity and Reaper doing straight subtraction. The null is less good being only -63 db instead of more than -80 db.
@pma @pkane

 

[pkane]

When I used Deltawave the difference file has a click when amplified at that 41.3 second point where the discontinuity is. When I do a straight subtraction in Audacity it is still there though more of sharp thump than a click. When I do this in Reaper it is a thump like in Audacity.

Now if I turn off "Correct clock drift" and "Subsample Offset" in Deltawave then I get the thump like in Audacity and Reaper doing straight subtraction. The null is less good being only -63 db instead of more than -80 db.
@pma @pkane

The "thump" is an improper signal in a 44.1k sampling rate, containing frequencies above Nyquist and causing unfiltered aliases. Here's what that discontinuity looks like when the whole recording is filtered with a 21.5k low pass filter (blue is the original showing original samples, red is filtered). You can clearly see Gibbs around the discontinuity when invalid frequencies are eliminated.

 

[pma]

When I used Deltawave the difference file has a click when amplified at that 41.3 second point where the discontinuity is. When I do a straight subtraction in Audacity it is still there though more of sharp thump than a click. When I do this in Reaper it is a thump like in Audacity.

Now if I turn off "Correct clock drift" and "Subsample Offset" in Deltawave then I get the thump like in Audacity and Reaper doing straight subtraction. The null is less good being only -63 db instead of more than -80 db.
@pma @pkane

Interesting. I am getting this from Adobe Audition, as a direct subtraction. Where do I see the discontinuity? But yes, I did not try to download the uploaded files, but I am checking my originals in the PC.

 

[pkane]

Interesting. I am getting this from Adobe Audition, as a direct subtraction. Where do I see the discontinuity? But yes, I did not try to download the uploaded files, but I am checking my originals in the PC.

View attachment 373326

View attachment 373327

View attachment 373328

It's obscured by the large error due to a coarse sample alignment. The RMS of the difference you show (when aligning on a sample boundary) is -63dB. This is what DW shows too, if you turn off sub-sample alignment option. With sub-sample alignment, DW null is -80.5dB or 17dB lower.

Aligning on a sample boundary doesn't produce a proper null since the error can be as much as 22usec for all the samples. With proper sub-sample alignment, DW computes an offset of 0.10384 of a sample and this makes a 17dB difference in the difference result.

 

[Blumlein 88]

Interesting. I am getting this from Adobe Audition, as a direct subtraction. Where do I see the discontinuity? But yes, I did not try to download the uploaded files, but I am checking my originals in the PC.

View attachment 373326

View attachment 373327

View attachment 373328

I think if you'll listen you'll hear it. It is not as easy to hear as the click, but is definitely there.

Deltawave shows just above a 1 microsecond offset between the files.