DeltaWave null-testing audio comparator (beta)

[TomCapraro]

2 minutes ago, pkane2001 said:

 

Can you please also post the text from Results tab? This has all the numbers with many more decimal places.

Of both versions? I should install 1.0.45 again

[pkane2001]

Just now, TomCapraro said:

Of both versions? I should install 1.0.45 again

 

 v1.0.46  will be enough, Tom.

[TomCapraro]

DeltaWave v1.0.46, 2019-11-05T18:43:22.1086502+01:00
Reference:  White_PN_64k_20_20000_96.0k_Float_LR.wav[L] 1966080 samples 96000Hz 32bits, stereo, MD5=00
Comparison: A REC WHITE PN64K.wav[L] 2030302 samples 96000Hz 32bits, mono, MD5=00
Settings: 
    Gain:True, Remove DC:True
    Non-linear Gain EQ:True    Non-linear Phase EQ: True
    EQ FFT Size:65536, EQ Frequency Cut: 0Hz - 20000Hz, EQ Threshold: -160dB
    Correct Drift:True, Precision:30
    Non-Linear drift Correction:True
    Upsample:False, Window:Hann
    Spectrum Window:Dirichlet, Spectrum Size:65536
    Spectrogram Window:Hann, Spectrogram Size:4096, Spectrogram Steps:2048
    Dither:False
    Trim Silence:True
    Enable Simple Waveform Measurement: True

Discarding Reference:  Start=0s, End=0s
Discarding Comparison: Start=0s, End=0s

Initial peak values Reference: -1,743dB   Comparison: -0,947dB
Initial RMS values Reference: -7,71dB   Comparison: -7,312dB

Null Depth=12,646dB
X-Correlation offset: -28332 samples
Trimming 0 samples at start and 1 samples at the end that are below -90,31dB level

Trimmed 0 samples ( 0,00ms) front, 0 samples ( 0,00ms end)

Final peak values Reference: -1,743dB   Comparison: -1,743dB
Final RMS values Reference: -7,711dB   Comparison: -7,711dB

Gain= 6,0132dB (1,9983x) DC=0 Phase offset=-295,124559ms (-28331,958 samples)
Difference (rms) = -79,55dB [-84,07dBA]
Correlated Null Depth=93,67dB [98,9dBA]
Clock drift: 0 ppm

Files are NOT a bit-perfect match (match=5,73%) at 16 bits
Files are NOT a bit-perfect match (match=0%) at 32 bits
Files match @ 49,9976% when reduced to 12,68 bits

---- Phase difference (full bandwidth): 6,22101875209607°
    0-10kHz: 0,58°
    0-20kHz: 0,41°
    0-24kHz: 2,45°
Phase EQ not computed
Timing error (rms jitter): 30,3ps

RMS of the difference of spectra: -135,543141040123dB
gn=0,500427029006524, dc=0, dr=0, of=-28331,9576750275

---Measurements (for a simple sine-wave only)---

Stopped! Indice oltre i limiti della matrice.
Signature: fe3bb3b450827d6f3b0c451043facd13

[TomCapraro]

I won't hide Paul that I would have liked it better if the same functions (including the group delay) as the 1.045 version remained but with the display of the distortion amplitude with the darts, which is appreciable😋

[pkane2001]

57 minutes ago, TomCapraro said:

I won't hide Paul that I would have liked it better if the same functions (including the group delay) as the 1.045 version remained but with the display of the distortion amplitude with the darts, which is appreciable😋

 

I don't have strong feelings about the group delay line. It was originally there because that's what I used to correct for group delay, but then changed the method to a new one that doesn't require approximation. I can certainly turn it back on

[TomCapraro]

Perhaps if you reset this it also works by reducing the jitter value and null-rms

 

Forgot one more new feature: In simple waveform measurement mode drift measurement and correction is now working as long as the drift is not overly excessive.

[pkane2001]

17 hours ago, TomCapraro said:

Perhaps if you reset this it also works by reducing the jitter value and null-rms

 

Forgot one more new feature: In simple waveform measurement mode drift measurement and correction is now working as long as the drift is not overly excessive.

 

Hi Tom, I don't think that the phase approximation plot had anything to do with the increased jitter number They are not related. What might help me is if you could please share with me the two files you used to measure the difference in jitter. Before turning off drift correction, I'd like to understand what really caused the difference.

 

Regards,

     -Paul

[TomCapraro]

47 minutes ago, pkane2001 said:

 

Hi Tom, I don't think that the phase approximation plot had anything to do with the increased jitter number They are not related. What might help me is if you could please share with me the two files you used to measure the difference in jitter. Before turning off drift correction, I'd like to understand what really caused the difference.

 

Regards,

     -Paul

Immediately...!!! here they are

White_PN_64k_20_20000_96.0k_Float_LR.wav A REC WHITE PN64K.wav

[TomCapraro]

I am "insisting" on this path because that result turns out to be congruous if compared to the real non-linear alterations of the dac.
The 1.045 version is the best from many points of view.
It is enough to preserve it and only add new features without affecting the excellent yield it has shown to have.

[Arpiben]

Hi @pkane2001

 

The phase drift correction is having some issues with simple waveforms at least.

 

1° There is a gain value of 4 applied to comparison once drift is selected. In order to get the proper estimation one must select Match Gain. I am suspecting there is an upsampling gain not corrected/taken into account.

 

2° Dealing with integer offset sample values lead to correct/accurate result only if phase drift is unselected. Otherwise we are getting wrong offset/drift values.

 

3° Dealing with subsample offset values can you tell what is the expected resolution?

 

Tests have been performed with a 1kHz pure sine with a fixed timed offset:

y = sin(2*pi*f0*(t+/-TIE)        f0=1kHz / fs=64kHz / TIE= k/fs or 1/(k*fs)

 

Rgds

[modmix]

On 11/5/2019 at 2:46 PM, pkane2001 said:

3) The method for non-linearity computes the error for each individual sample value. This is repeated for all possible sample values in the file and plotted on a chart.  Any sample values that occur rarely in the file will have a less accurate error value. White noise as close as possible to 0dBFS max value is probably best for measuring non-linearity, as it should produce an equal number of all sample values in the file.

I wonder what's the best setting to get a good result.

 

A 24bit white noise file is played via Lynx L22 digital out connected to a Soekris DAC1541. The DAC1541 output is recorded from attached L22 analog input (loop with Adobe Audition 3.0).

 

A 20 - 40 dB wide "band" is not what I would expect.

 

Thanks,
Ulli

 

BTW: any chance to allow for export e.g. as .csv?

 

[TomCapraro]

33 minutes ago, modmix said:

I wonder what's the best setting to get a good result.

 

A 24bit white noise file is played via Lynx L22 digital out connected to a Soekris DAC1541. The DAC1541 output is recorded from attached L22 analog input (loop with Adobe Audition 3.0).

 

A 20 - 40 dB wide "band" is not what I would expect.

 

Thanks,
Ulli

 

BTW: any chance to allow for export e.g. as .csv?

 

Download the White_PN_64k_20_20000_96.0k_Float_LR file I posted, record it and analyze it exactly with the same settings I provided.

[modmix]

Quite the same, I'ld say.
What was your intention?

[pkane2001]

4 hours ago, modmix said:

I wonder what's the best setting to get a good result.

 

A 24bit white noise file is played via Lynx L22 digital out connected to a Soekris DAC1541. The DAC1541 output is recorded from attached L22 analog input (loop with Adobe Audition 3.0).

 

A 20 - 40 dB wide "band" is not what I would expect.

 

Thanks,
Ulli

 

BTW: any chance to allow for export e.g. as .csv?

 

 

Hi Ulli,

 

The null must be very good for non-linearity errors to be close to reality. Looking at your results, the null is poor, so the errors per individual sample value will be very large. With the non-linear EQ for phase and level, I get better than -70dB null

 

Also, I recommend that you trim the files (turn on auto-trim option) as that can throw computations off if there are samples that are completely mismatched at the start or end of the recording.

 

Also try larger FFT size and Kaiser window for Spectrogram and Non-linear EQ settings. See if the results improve.

 

[pkane2001]

11 hours ago, Arpiben said:

Hi @pkane2001

 

1° There is a gain value of 4 applied to comparison once drift is selected. In order to get the proper estimation one must select Match Gain. I am suspecting there is an upsampling gain not corrected/taken into account.

 

Which window and FFT size? A 4x gain is very strange and shouldn't happen.

 

I'll try to reproduce this here.

[pkane2001]

Removed -- need more research

[pkane2001]

3 hours ago, modmix said:

Quite the same, I'ld say.
What was your intention?

 

Ulli, here's what I get with the two files that Tom posted for non-linearity. Something doesn't seem right with how you are either recording or matching the two files:

[pkane2001]

12 hours ago, TomCapraro said:

Immediately...!!! here they are

White_PN_64k_20_20000_96.0k_Float_LR.wav 15 MB · 2 downloads A REC WHITE PN64K.wav 7.75 MB · 2 downloads

 

Tom, if I pick a larger FFT size, I get a much better result than in v1.0.45. Better null value and better jitter value.

 

Can you please try the same settings?

 

 

[pkane2001]

37 minutes ago, pkane2001 said:

 

Which window and FFT size? A 4x gain is very strange and shouldn't happen.

 

I'll try to reproduce this here.

 

@Arpiben, you mentioned resampling? What were the two sampling rates, and what was the resampling option in settings?

[Arpiben]

3 hours ago, pkane2001 said:

 

@Arpiben, you mentioned resampling? What were the two sampling rates, and what was the resampling option in settings?

 

Hi Paul,

 

It is happening with whatever window at least with joined files: no resampling and 64kHz rate as previously mentioned.

It is happening only with simple waveform mode and phase drift correction as soon as you press match.

Rgds.

 

 

 

A.wav B.wav

[TomCapraro]

10 hours ago, pkane2001 said:

 

Tom, if I pick a larger FFT size, I get a much better result than in v1.0.45. Better null value and better jitter value.

 

Can you please try the same settings?

 

 

The 1.045 version generates 7.6ps with the settings you recommended

[pkane2001]

Just now, TomCapraro said:

The 1.045 version generates 7.6ps with the settings you recommended

 

Even better!  That means v1.0.46 made jitter calculation worse for all FFT sizes. Let me try to find out why.

 

[pkane2001]

14 hours ago, modmix said:

Quite the same, I'ld say.
What was your intention?

 

Hi Ulli,

 

Thanks for sending me your recording. I see the problem when using your exact settings. Strangely, when I skip the first couple of seconds, the result with your recording is much, much better (both, null and non-linearity values):

 

I'll need to explore more to understand why. I suspect this might be related to the gain-matching issue that @Arpiben reported.

 

 

[Arpiben]

8 hours ago, Arpiben said:

 

Hi Paul,

 

It is happening with whatever window at least with joined files: no resampling and 64kHz rate as previously mentioned.

It is happening only with simple waveform mode and phase drift correction as soon as you press match.

Rgds.

 

 

 

A.wav 4 MB · 0 downloads B.wav 4 MB · 0 downloads

 Hi @pkane2001

 

In case it helps the gain=4 applied to Comparison is also present with whatever audio files/FFT length/window provided that:

Correct Phase Drift + Measure Simple Waveform + NO Match Gain are enabled. 

Rgds

 

N.B. if Match Gain is enabled then Comparison will have a correction of something like gn = 0.25*****.

 

[lucretius]

Hi Paul,

 

When I close DeltaWave, I get this error message:

 

 

Any ideas?  Thanks!