[bfreedma]

  Like I said, archimago made poor measurements that audioasylum users don't accept his tests. I already provided a link in this post so go read the thread yourself.
 
http://www.head-fi.org/t/795259/fidelizer-pro-real-or-snake-oil#post_12265629
 
As for why it's so bad, check MindsMirror's post here. Archimago may have better gears but it hardly tells anything comparing to this.
 
http://www.head-fi.org/t/795483/audio-optimizations-on-bit-perfect-playback-demonstration/45#post_12279759
 
And I'm not playing with watchnerd. He seriously needs some mental aid. Normal people would not spent all time of the day posting and arguing up to the point that all people who argued with him left except me who do it in spare time.
 
Regards,
Windows X

 
Followed by this request to Watchnerd:
 

   
That's why I need to ask you what digital metrics do you accept for measurements so I can conduct the tests accordingly.
 
Regards,
Windows X

 
 
Bolded parts reposted just to highlight how ridiculous this discussion has become.

 

[WindowsX]

Yes. It's really ridiculous now. Are you done digging old posts? Since you don't agree about using analog domain metric on digitally  recorded files, I'll propose my approach now.
 
1. I'll setup master file and audio playback/recording through digital domain. In this case, I'll use VB-Audio Virtual Cable, foobar2000, and Audacity on Windows 10.
2. I'll use VB-Audio Virtual Cable to route master audio stream from Foobar2000's WASAPI output to Audacity's WASAPI input, export audio as normal.wav
3. I'll use Fidelizer and record again without turning off software, export audio as fidelizer.wav
4. I'll use record again one more time, export audio as fidelizer-again.wav
5. I'll open Audio DiffMaker, align all recording to master track, then make difference tracks.
6. I'll pass the test result to you guys for analysis.
 
How does this sound?
 
Regards,
Windows X

 

[watchnerd]

   Since you don't agree about using analog domain metric on digitally  recorded files

 
ROFLMFAO!!!
 
The reason I don't agree is because it's not possible!
 
You can't measure THD, IMD, channel-separation, etc in the digital domain.  They don't exist until you convert to analog!
 
The fact that you didn't know this shows just how little you know about audio engineering.

 

[WindowsX]

   
ROFLMFAO!!!
 
The reason I don't agree is because it's not possible!
 
You can't measure THD, IMD, channel-separation, etc in the digital domain.  They don't exist until you convert to analog!
 
The fact that you didn't know this shows just how little you know about audio engineering.

 
I already told you that measuring THD, IMD requires DAC and it has analog output stage, digital filters and stuff that can affect changes in digital domain. I did know this and I want to propose such method and measurements without DA conversion. What do you think about DiffMaker?
 
Regards,
Windows X

 

[watchnerd]

 
I did know this and I want to propose such method and measurements without DA conversion.

 
No, you're not getting it...you can't take those measurements without DA conversion.  It doesn't matter if it's a soundcard or an external DAC.
 
They don't exist to be measured....

 

[WindowsX]

   
No, you're not getting it...you can't take those measurements without DA conversion.  It doesn't matter if it's a soundcard or an external DAC.
 
They don't exist to be measured....

 
You can bypass DA conversion by recording from virtual audio sound card. Just route from output to input virtually without real audio interface, DA conversion, pure software. You can see this study case for an example.
 
http://vb-audio.pagesperso-orange.fr/Cable/VBCABLE_CaseStudy_XP.pdf
 
Measuring between WAV files is possible whether you'll accept the data from digital recording or not.
 
Regards,
Windows X

 

[watchnerd]

   
You can bypass DA conversion by recording from virtual audio sound card. Just route from output to input virtually without real audio interface, DA conversion, pure software. You can see this study case for an example.
 
http://vb-audio.pagesperso-orange.fr/Cable/VBCABLE_CaseStudy_XP.pdf
 
Measuring between WAV files is possible whether you'll accept the data from digital recording or not.
 
Regards,
Windows X

 
Right...you're doing digital-to-digital.
 
Which means you can't take any real-time measurements of THD, IMD, etc.  All you can test-real time is digital domain stuff:  jitter, sample rate, bit depth, etc.
 
You can try to use DiffMaker afterwards, but the differences should be even less than what you see in the analog domain (the tests you didn't like).
 
That being said, go for it. I'm curious to see what you get.

 

[nick_charles]

  Yes. It's really ridiculous now. Are you done digging old posts? Since you don't agree about using analog domain metric on digitally  recorded files, I'll propose my approach now.
 
1. I'll setup master file and audio playback/recording through digital domain. In this case, I'll use VB-Audio Virtual Cable, foobar2000, and Audacity.
2. I'll use VB-Audio Virtual Cable to route master audio stream from Foobar2000's WASAPI output to Audacity's WASAPI input, export audio as normal.wav
3. I'll use Fidelizer and record again without turning off software, export audio as fidelizer.wav
4. I'll use record again one more time, export audio as fidelizer-again.wav
5. I'll open Audio DiffMaker, align all recording to master track, then make difference tracks.
6. I'll pass the test result to you guys for analysis.
 
How does this sound?
 
Regards,
Windows X

 
 
Before doing step 5 make sure DiffMaker works on your system - I've had **several** versions in the past which could detect a difference between identical files!

 

[watchnerd]

   
 
Before doing step 5 make sure DiffMaker works on your system - I've had **several** versions in the past which could detect a difference between identical files!

 
how many dB down were those differences on identical files?

 

[WindowsX]

   
Right...you're doing digital-to-digital.
 
Which means you can't take any real-time measurements of THD, IMD, etc.  All you can test-real time is digital domain stuff:  jitter, sample rate, bit depth, etc.
 
You can try to use DiffMaker afterwards, but the differences should be even less than what you see in the analog domain (the tests you didn't like).
 
That being said, go for it. I'm curious to see what you get.

 
Real-time measurements might possible with closed loop virtual audio interface. I haven't tested it yet so I'll start with wav files comparison first. That way, you can check on test result files yourself instead of reading my data alone.
 

   
 
Before doing step 5 make sure DiffMaker works on your system - I've had **several** versions in the past which could detect a difference between identical files!

 
I see. I'll test this case with master file first. Thank you for informing me this.
 
Regards,
Windows X

 

[WindowsX]

   
how many dB down were those differences on identical files?

 
I tried comparing between the same file and got this result
 
"parameters: 0sec, 0.000dB (L),  0.000dB (R)..Corr Depth: 300.0 dB (L), 300.0 dB (R)"
 
However, I created master-test@master and master-test-master with smaller size (no id tag cover) but same number of samples. Is it safe?
 
Regards,
Windows X

 

[nick_charles]

   
how many dB down were those differences on identical files?

 
Cant remember but there was an audible difference file

 

[WindowsX]

   
Cant remember but there was an audible difference file

 
Maybe it was created regardless of being the same or not? It didn't detect any difference from result besides creating test files.

 

[Don Hills]

   
Maybe it was created regardless of being the same or not? It didn't detect any difference from result besides creating test files.

It's probably working OK for you then.
To avoid problems, you could simply make the captured WAV files available instead of the diffs.
 
I predict you'll find a null result, provided you don't screw anything up. Your test method is equivalent to using a "perfect" DAC, with zero distortion and zero susceptibility to jitter and noise on the bitstream from the PC to the DAC. Perfect DACs don't need PC optimisers or USB isolators etc.

 

[WindowsX]

 
It's probably working OK for you then.
To avoid problems, you could simply make the captured WAV files available instead of the diffs.
 
I predict you'll find a null result, provided you don't screw anything up. Your test method is equivalent to using a "perfect" DAC, with zero distortion and zero susceptibility to jitter and noise on the bitstream from the PC to the DAC. Perfect DACs don't need PC optimisers or USB isolators etc.

 
That's correct. And it also avoids problems with real audio hardware interfacing. It's pure software experiment. However, this doesn't cover problems related to interfacing with real audio hardware but we'll leave that for now.
 
OK. I finished the experiment using the following setup
 
OS: Windows 10 (No Fidelizer Pro installed)
Sound Card: VB-Audio Virtual Cable (Max Latency: 1024smp)
Playback: Foobar2000 with no ReplayGain, Kernel Streaming output, 50ms Buffer length, Playback optimized
Record: Audacity with Windows WASAPI as Audio Host, 50ms Audio buffer
Test track: Astrognosia: Aquarius – Vannmannen (CD-resolution) from 2L High Resolution Music .:. free TEST BENCH
 
Procedures
 
1. Record before Fidelizer optimizations
2. Use Fidelizer with Extremist optimization level (no services to keep), exit all programs
3. Record after Fidelizer optimizations
4. Use DiffMaker to compare each recording with reference file from test bench
5. Compare between recordings (untouched with different samples)
6. Compare between aligned recordings (same samples)
7. Listen difference track for audible sound
 
Test Results
 
1. I recorded before and after files without issues. No stuttering and I'll use before recording as main file so after will be slightly longer.
 
2. I used DiffMaker to compare before file with master recording and got this result
 

 parameters: -1.596sec, 0.001dB (L),  0.002dB (R)..Corr Depth: 79.2 dB (L), 72.2 dB (R)

 
0.00xdB is still in marginal error and inaudible. I consider this as null result. I tried my best to optimize software and driver before recording but still can't get 0.000dB. Moving on
 
3. I used DiffMaker to compare after file with master recording and got this result
 

parameters: -1.852sec, 0.002dB (L),  0.003dB (R)..Corr Depth: 73.8 dB (L), 69.5 dB (R)

 
0.00xdB is still in marginal error and inaudible. I consider this as null result. I noticed 0.001dB increment but that is also marginal error from alignment. Moving on.
 
4. I used DiffMaker to compare reference file before with after file and got this result
 

parameters: -255.9msec, 0.136dB (L),  0.136dB (R)..Corr Depth: 36.0 dB (L), 36.0 dB (R)

 
Exactly 0.136dB for both channel? This is beyond marginal error and should be audible. This can't be null result. Maybe changes between recording has long delay to cause something like this. Let's try aligned tracks
 
5. I used DiffMaker to compare reference file before with after file using aligned source and got this result
 

parameters: 0sec, 0.026dB (L),  0.012dB (R)..Corr Depth: 50.1 dB (L), 56.3 dB (R)

 
Looks much better but it's still beyond marginal error. Some parts should be audible. This can't be null result. Maybe different recording has this level of marginal error?
 
6. I used DiffMaker to compare reference file before with before_again file using aligned source and got this result
 

parameters: 0sec, 0.027dB (L),  0.013dB (R)..Corr Depth: 49.3 dB (L), 55.6 dB (R)

 
Looks like marginal error between recording. And Left is twice loader after alignment. I'm using bit-perfect in the chain for both playback (Kernel Streaming) and recording (WASAPI). How come there's still such big marginal error like this? It looks fine with master file as reference, though.

 
7. I tried listening to difference tracks, I can't hear anything when use master file as reference but can hear faint sound from between recording files regardless of using Fidelizer optimizations or not.
 
Conclusion
 
Bit-perfect playback/recording can produce marginal error at range 0.01-0.03dB. Still need to investigate to make null result between recording. If you have any good freeware or configuration to recommend as alternatives, please post below and I'll check again tomorrow.
 
Regards,
Windows X