DeltaWave null-testing audio comparator (beta)

[pkane2001]

6 minutes ago, fas42 said:

Paul, haven't used DW for a bit - and installing the new one has brought up some peculiarities. Most notably, not getting most displays in Tabs to present! 🤪 Got the loss of permissions problem that @lucretius got, which may have something to do with it, but fixing that has not helped.

 

About to go to bed, so may have missed something obvious - try again in the morning, 🙂.

Hi Frank,

 

Please try to re-download and install the version from the site. There was a version I posted earlier that had some of the memory management logic that could interfere with charts display (it would stop charts from populating when detecting that there’s not enough memory left). That’s been fixed since yesterday.

[fas42]

Yep, that sorted it ... thanks!

 

 

[TomCapraro]

2020-10-16 16:57:13.9211|INFO|Wave.WaveForm|DeltaWave 1.0.51.0 starting up
2020-10-16 16:57:13.9836|INFO|Wave.WaveForm|Adding driver: [ASIO] JRiver Media Center 20
2020-10-16 16:57:13.9836|INFO|Wave.WaveForm|Adding driver: [ASIO] Realtek ASIO
2020-10-16 16:57:14.0460|INFO|Wave.WaveForm|Adding driver: [WASAPI]{0.0.0.00000000}.{4329faf5-84fb-470b-ad84-4e9d9ba44bf3} | [WASAPI] Realtek Digital Output (Realtek High Definition Audio) 96000/32
2020-10-16 16:57:14.0929|INFO|Wave.WaveForm|Adding driver: [WASAPI]{0.0.0.00000000}.{abdcf97b-55e5-47fd-a4eb-791c4f66db3e} | [WASAPI] Altoparlanti (Realtek High Definition Audio) 44100/32
2020-10-16 16:57:14.0929|INFO|Wave.WaveForm|Current driver: [WASAPI]{0.0.0.00000000}.{abdcf97b-55e5-47fd-a4eb-791c4f66db3e}
2020-10-16 16:57:14.1085|INFO|Wave.WaveForm|Adding driver: [ASIO] JRiver Media Center 20
2020-10-16 16:57:14.1085|INFO|Wave.WaveForm|Adding driver: [ASIO] Realtek ASIO
2020-10-16 16:57:14.1554|INFO|Wave.WaveForm|Adding driver: [WASAPI]{0.0.0.00000000}.{4329faf5-84fb-470b-ad84-4e9d9ba44bf3} | [WASAPI] Realtek Digital Output (Realtek High Definition Audio) 96000/32
2020-10-16 16:57:14.2023|INFO|Wave.WaveForm|Adding driver: [WASAPI]{0.0.0.00000000}.{abdcf97b-55e5-47fd-a4eb-791c4f66db3e} | [WASAPI] Altoparlanti (Realtek High Definition Audio) 44100/32
2020-10-16 16:57:14.2023|INFO|Wave.WaveForm|Current driver: [WASAPI]{0.0.0.00000000}.{abdcf97b-55e5-47fd-a4eb-791c4f66db3e}
2020-10-16 16:59:49.2909|INFO|Wave.WaveForm|ProcessAll
2020-10-16 16:59:49.5252|INFO|Wave.WaveForm|DeltaWave v1.0.51, 2020-10-16T16:59:49.5252304+02:00
2020-10-16 16:59:49.5252|INFO|Wave.WaveForm|Reference:  White_PN_64k_20_20000_96.0k_Float_LR.wav[L] 1966080 samples 96000Hz 32bits, stereo, MD5=00
2020-10-16 16:59:49.5252|INFO|Wave.WaveForm|Comparison: rec 1.wav[L] 2039902 samples 96000Hz 32bits, mono, MD5=00
2020-10-16 16:59:49.5252|INFO|Wave.WaveForm|Settings: 
    Gain:True, Remove DC:True
    Non-linear Gain EQ:True    Non-linear Phase EQ: True
    EQ FFT Size:65536, EQ Frequency Cut: 0Hz - 0Hz, EQ Threshold: -96dB
    Correct Drift:True, Precision:30
    Non-Linear drift Correction:False
    Upsample:False, Window:Dirichlet
    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

2020-10-16 16:59:49.5252|INFO|Wave.WaveForm|Discarding Reference:  Start=0s, End=0s
2020-10-16 16:59:49.5252|INFO|Wave.WaveForm|Discarding Comparison: Start=0s, End=0s
2020-10-16 16:59:49.7439|INFO|Wave.WaveForm|
Initial peak values Reference: -1,743dB   Comparison: -1,455dB
2020-10-16 16:59:49.7439|INFO|Wave.WaveForm|Initial RMS values Reference: -7,71dB   Comparison: -7,847dB

2020-10-16 16:59:49.7595|INFO|Wave.WaveForm|Null Depth=23,173dB
2020-10-16 16:59:55.3832|INFO|Wave.WaveForm|Stopped! A regression of the requested order requires at least 2 samples. Only 1 samples have been provided. 
2020-10-16 16:59:55.3832|INFO|Wave.WaveForm|Signature: 897074275db140d8a51412b4daa266b0
2020-10-16 17:01:48.7785|INFO|Wave.WaveForm|DeltaWave 1.0.51.0 starting up
2020-10-16 17:01:48.8409|INFO|Wave.WaveForm|Adding driver: [ASIO] JRiver Media Center 20
2020-10-16 17:01:48.8409|INFO|Wave.WaveForm|Adding driver: [ASIO] Realtek ASIO
2020-10-16 17:01:48.9034|INFO|Wave.WaveForm|Adding driver: [WASAPI]{0.0.0.00000000}.{4329faf5-84fb-470b-ad84-4e9d9ba44bf3} | [WASAPI] Realtek Digital Output (Realtek High Definition Audio) 96000/32
2020-10-16 17:01:48.9503|INFO|Wave.WaveForm|Adding driver: [WASAPI]{0.0.0.00000000}.{abdcf97b-55e5-47fd-a4eb-791c4f66db3e} | [WASAPI] Altoparlanti (Realtek High Definition Audio) 44100/32
2020-10-16 17:01:48.9503|INFO|Wave.WaveForm|Current driver: [WASAPI]{0.0.0.00000000}.{abdcf97b-55e5-47fd-a4eb-791c4f66db3e}
2020-10-16 17:01:49.0128|INFO|Wave.WaveForm|Adding driver: [ASIO] JRiver Media Center 20
2020-10-16 17:01:49.0128|INFO|Wave.WaveForm|Adding driver: [ASIO] Realtek ASIO
2020-10-16 17:01:49.0596|INFO|Wave.WaveForm|Adding driver: [WASAPI]{0.0.0.00000000}.{4329faf5-84fb-470b-ad84-4e9d9ba44bf3} | [WASAPI] Realtek Digital Output (Realtek High Definition Audio) 96000/32
2020-10-16 17:01:49.1065|INFO|Wave.WaveForm|Adding driver: [WASAPI]{0.0.0.00000000}.{abdcf97b-55e5-47fd-a4eb-791c4f66db3e} | [WASAPI] Altoparlanti (Realtek High Definition Audio) 44100/32
2020-10-16 17:01:49.1065|INFO|Wave.WaveForm|Current driver: [WASAPI]{0.0.0.00000000}.{abdcf97b-55e5-47fd-a4eb-791c4f66db3e}
2020-10-16 17:02:28.1775|INFO|Wave.WaveForm|ProcessAll
2020-10-16 17:02:28.3025|INFO|Wave.WaveForm|DeltaWave v1.0.51, 2020-10-16T17:02:28.3025096+02:00
2020-10-16 17:02:28.3025|INFO|Wave.WaveForm|Reference:  White_PN_64k_20_20000_96.0k_Float_LR.wav[L] 1966080 samples 96000Hz 32bits, stereo, MD5=00
2020-10-16 17:02:28.3025|INFO|Wave.WaveForm|Comparison: rec 1.wav[L] 2039902 samples 96000Hz 32bits, mono, MD5=00
2020-10-16 17:02:28.3181|INFO|Wave.WaveForm|Settings: 
    Gain:True, Remove DC:True
    Non-linear Gain EQ:True    Non-linear Phase EQ: True
    EQ FFT Size:65536, EQ Frequency Cut: 0Hz - 48000Hz, EQ Threshold: -96dB
    Correct Drift:True, Precision:30
    Non-Linear drift Correction:False
    Upsample:False, Window:Dirichlet
    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

2020-10-16 17:02:28.3181|INFO|Wave.WaveForm|Discarding Reference:  Start=0s, End=0s
2020-10-16 17:02:28.3181|INFO|Wave.WaveForm|Discarding Comparison: Start=0s, End=0s
2020-10-16 17:02:28.5368|INFO|Wave.WaveForm|
Initial peak values Reference: -1,743dB   Comparison: -1,455dB
2020-10-16 17:02:28.5368|INFO|Wave.WaveForm|Initial RMS values Reference: -7,71dB   Comparison: -7,847dB

2020-10-16 17:02:28.5524|INFO|Wave.WaveForm|Null Depth=23,173dB
2020-10-16 17:02:34.0667|ERROR|Wave.WaveForm|Stopped!
   in MathNet.Numerics.LinearRegression.SimpleRegression.Fit(Double[] x, Double[] y)
   in Wave.Analysis.isPhaseInverted(Double[] L, Double[] L1, Int32 freq, Int32 data_length, Double& start_phs, Double& start_inv, Boolean& large_error, IInterpolation& drift) in C:\Users\ypa\documents\visual studio 2015\Projects\Wave\Wave\Analysis.cs:riga 195
   in Wave.WaveForm.Xcorrelate(Double[]& L, Double[]& L1, Int32 freq, Int32 bits1, Int32 bits2, Boolean bMatch, Boolean bApplyManual, Boolean bUpdateCharts, Double& offset, Double& start, Double[]& R, Double[]& R1, Boolean& large_error, Double mx, Double& drift) in C:\Users\ypa\documents\visual studio 2015\Projects\Wave\Wave\WaveForm.cs:riga 1714
   in Wave.WaveForm.ProcessAll(Double[] L, Double[] L1, Int32 freq, Int32 freq1, Int32 freq2, Int32 bits1, Int32 bits2, Boolean bMatch, Boolean bLoadOnly, Boolean bApplyManual, Boolean bUpdateCharts, Boolean bStereo) in C:\Users\ypa\documents\visual studio 2015\Projects\Wave\Wave\WaveForm.cs:riga 980
2020-10-16 17:02:34.0667|INFO|Wave.WaveForm|Stopped! A regression of the requested order requires at least 2 samples. Only 1 samples have been provided. 
2020-10-16 17:02:34.0667|INFO|Wave.WaveForm|Signature: 64d9f2371b4760148d89635e5788a48b
2020-10-16 17:02:48.1572|INFO|Wave.WaveForm|ProcessAll
2020-10-16 17:02:48.3290|INFO|Wave.WaveForm|DeltaWave v1.0.51, 2020-10-16T17:02:48.3290901+02:00
2020-10-16 17:02:48.3290|INFO|Wave.WaveForm|Reference:  rec 2.wav[L] 2039902 samples 96000Hz 32bits, mono, MD5=00
2020-10-16 17:02:48.3290|INFO|Wave.WaveForm|Comparison: rec 1.wav[L] 2039902 samples 96000Hz 32bits, mono, MD5=00
2020-10-16 17:02:48.3446|INFO|Wave.WaveForm|Settings: 
    Gain:True, Remove DC:True
    Non-linear Gain EQ:True    Non-linear Phase EQ: True
    EQ FFT Size:65536, EQ Frequency Cut: 0Hz - 48000Hz, EQ Threshold: -96dB
    Correct Drift:True, Precision:30
    Non-Linear drift Correction:False
    Upsample:False, Window:Dirichlet
    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

2020-10-16 17:02:48.3446|INFO|Wave.WaveForm|Discarding Reference:  Start=0s, End=0s
2020-10-16 17:02:48.3446|INFO|Wave.WaveForm|Discarding Comparison: Start=0s, End=0s
2020-10-16 17:02:48.5477|INFO|Wave.WaveForm|
Initial peak values Reference: -1,456dB   Comparison: -1,455dB
2020-10-16 17:02:48.5477|INFO|Wave.WaveForm|Initial RMS values Reference: -7,847dB   Comparison: -7,847dB

2020-10-16 17:02:48.5477|INFO|Wave.WaveForm|Null Depth=13,865dB
2020-10-16 17:02:48.6883|ERROR|Wave.WaveForm|Stopped!
   in MathNet.Numerics.LinearRegression.SimpleRegression.Fit(Double[] x, Double[] y)
   in Wave.Analysis.isPhaseInverted(Double[] L, Double[] L1, Int32 freq, Int32 data_length, Double& start_phs, Double& start_inv, Boolean& large_error, IInterpolation& drift) in C:\Users\ypa\documents\visual studio 2015\Projects\Wave\Wave\Analysis.cs:riga 195
   in Wave.WaveForm.Xcorrelate(Double[]& L, Double[]& L1, Int32 freq, Int32 bits1, Int32 bits2, Boolean bMatch, Boolean bApplyManual, Boolean bUpdateCharts, Double& offset, Double& start, Double[]& R, Double[]& R1, Boolean& large_error, Double mx, Double& drift) in C:\Users\ypa\documents\visual studio 2015\Projects\Wave\Wave\WaveForm.cs:riga 1714
   in Wave.WaveForm.ProcessAll(Double[] L, Double[] L1, Int32 freq, Int32 freq1, Int32 freq2, Int32 bits1, Int32 bits2, Boolean bMatch, Boolean bLoadOnly, Boolean bApplyManual, Boolean bUpdateCharts, Boolean bStereo) in C:\Users\ypa\documents\visual studio 2015\Projects\Wave\Wave\WaveForm.cs:riga 980
2020-10-16 17:02:49.4069|INFO|Wave.WaveForm|Stopped! A regression of the requested order requires at least 2 samples. Only 1 samples have been provided. 
2020-10-16 17:02:49.4069|INFO|Wave.WaveForm|Signature: 08c6fa4efcc2aaa38505fc28be6a0e81
2020-10-16 17:05:00.6071|INFO|Wave.WaveForm|DeltaWave 1.0.51.0 starting up
2020-10-16 17:05:00.6696|INFO|Wave.WaveForm|Adding driver: [ASIO] JRiver Media Center 20
2020-10-16 17:05:00.6696|INFO|Wave.WaveForm|Adding driver: [ASIO] Realtek ASIO
2020-10-16 17:05:00.7321|INFO|Wave.WaveForm|Adding driver: [WASAPI]{0.0.0.00000000}.{4329faf5-84fb-470b-ad84-4e9d9ba44bf3} | [WASAPI] Realtek Digital Output (Realtek High Definition Audio) 96000/32
2020-10-16 17:05:00.7789|INFO|Wave.WaveForm|Adding driver: [WASAPI]{0.0.0.00000000}.{abdcf97b-55e5-47fd-a4eb-791c4f66db3e} | [WASAPI] Altoparlanti (Realtek High Definition Audio) 44100/32
2020-10-16 17:05:00.7789|INFO|Wave.WaveForm|Current driver: [WASAPI]{0.0.0.00000000}.{abdcf97b-55e5-47fd-a4eb-791c4f66db3e}
2020-10-16 17:05:00.8415|INFO|Wave.WaveForm|Adding driver: [ASIO] JRiver Media Center 20
2020-10-16 17:05:00.8415|INFO|Wave.WaveForm|Adding driver: [ASIO] Realtek ASIO
2020-10-16 17:05:00.8883|INFO|Wave.WaveForm|Adding driver: [WASAPI]{0.0.0.00000000}.{4329faf5-84fb-470b-ad84-4e9d9ba44bf3} | [WASAPI] Realtek Digital Output (Realtek High Definition Audio) 96000/32
2020-10-16 17:05:00.9352|INFO|Wave.WaveForm|Adding driver: [WASAPI]{0.0.0.00000000}.{abdcf97b-55e5-47fd-a4eb-791c4f66db3e} | [WASAPI] Altoparlanti (Realtek High Definition Audio) 44100/32
2020-10-16 17:05:00.9352|INFO|Wave.WaveForm|Current driver: [WASAPI]{0.0.0.00000000}.{abdcf97b-55e5-47fd-a4eb-791c4f66db3e}
2020-10-16 17:05:38.8325|INFO|Wave.WaveForm|ProcessAll
2020-10-16 17:05:39.0200|INFO|Wave.WaveForm|DeltaWave v1.0.51, 2020-10-16T17:05:39.0200497+02:00
2020-10-16 17:05:39.0200|INFO|Wave.WaveForm|Reference:  rec 1.wav[L] 2039902 samples 96000Hz 32bits, mono, MD5=00
2020-10-16 17:05:39.0200|INFO|Wave.WaveForm|Comparison: rec 2.wav[L] 2039902 samples 96000Hz 32bits, mono, MD5=00
2020-10-16 17:05:39.0200|INFO|Wave.WaveForm|Settings: 
    Gain:True, Remove DC:True
    Non-linear Gain EQ:True    Non-linear Phase EQ: True
    EQ FFT Size:65536, EQ Frequency Cut: 0Hz - 48000Hz, EQ Threshold: -96dB
    Correct Drift:True, Precision:30
    Non-Linear drift Correction:False
    Upsample:False, Window:Dirichlet
    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

2020-10-16 17:05:39.0200|INFO|Wave.WaveForm|Discarding Reference:  Start=0s, End=0s
2020-10-16 17:05:39.0200|INFO|Wave.WaveForm|Discarding Comparison: Start=0s, End=0s
2020-10-16 17:05:39.2543|INFO|Wave.WaveForm|
Initial peak values Reference: -1,455dB   Comparison: -1,456dB
2020-10-16 17:05:39.2543|INFO|Wave.WaveForm|Initial RMS values Reference: -7,847dB   Comparison: -7,847dB

2020-10-16 17:05:39.2543|INFO|Wave.WaveForm|Null Depth=13,861dB
2020-10-16 17:05:44.8467|ERROR|Wave.WaveForm|Stopped!
   in MathNet.Numerics.LinearRegression.SimpleRegression.Fit(Double[] x, Double[] y)
   in Wave.Analysis.isPhaseInverted(Double[] L, Double[] L1, Int32 freq, Int32 data_length, Double& start_phs, Double& start_inv, Boolean& large_error, IInterpolation& drift) in C:\Users\ypa\documents\visual studio 2015\Projects\Wave\Wave\Analysis.cs:riga 195
   in Wave.WaveForm.Xcorrelate(Double[]& L, Double[]& L1, Int32 freq, Int32 bits1, Int32 bits2, Boolean bMatch, Boolean bApplyManual, Boolean bUpdateCharts, Double& offset, Double& start, Double[]& R, Double[]& R1, Boolean& large_error, Double mx, Double& drift) in C:\Users\ypa\documents\visual studio 2015\Projects\Wave\Wave\WaveForm.cs:riga 1714
   in Wave.WaveForm.ProcessAll(Double[] L, Double[] L1, Int32 freq, Int32 freq1, Int32 freq2, Int32 bits1, Int32 bits2, Boolean bMatch, Boolean bLoadOnly, Boolean bApplyManual, Boolean bUpdateCharts, Boolean bStereo) in C:\Users\ypa\documents\visual studio 2015\Projects\Wave\Wave\WaveForm.cs:riga 980
2020-10-16 17:05:44.8467|INFO|Wave.WaveForm|Stopped! A regression of the requested order requires at least 2 samples. Only 1 samples have been provided. 
2020-10-16 17:05:44.8467|INFO|Wave.WaveForm|Signature: c3c398c1cb0aadedb30140a61d438ec4
 

[TomCapraro]

Hi Paul, something is wrong.
The calculation no longer starts.
p.s what is "PK METRIC" ???

[pkane2001]

39 minutes ago, TomCapraro said:

Hi Paul, something is wrong.
The calculation no longer starts.
p.s what is "PK METRIC" ???

 

Hi Tom,

 

Calculation no longer starts? What happens when you click Match? Is there an error message?

 

Ah, PK Metric was just a different approach to DF Metric, applying an equal loudness curve to the error data. It's an experiment, and probably not worth using just yet.

 

[TomCapraro]

No Paul, no error messages.
It stops after a few seconds without providing any data.
If, on the other hand, I reset with the default settings, it manages to complete the calculation but incorrect graphics are generated.

[TomCapraro]

I'll post the settings and comparison files so you can try.

White_PN_64k_20_20000_96.0k_Float_LR.wav rec 1.wav

[pkane2001]

 

19 minutes ago, TomCapraro said:

I'll post the settings and comparison files so you can try.

White_PN_64k_20_20000_96.0k_Float_LR.wav 15 MB · 0 downloads rec 1.wav 7.78 MB · 0 downloads

 

Can you please try to uncheck "Measure simple waveforms"? That's designed for periodic signals that have many repeating periods within a single capture. White noise is not very periodic ;)

 

For simple waveforms, there is a problem that I'll need to fix in this version. I'll post an update a bit later today.

 

[TomCapraro]

7 minutes ago, pkane2001 said:

 

 

Can you please try to uncheck "Measure simple waveforms"? That's designed for periodic signals that have many repeating periods within a single capture. White noise is not very periodic ;)

 

For simple waveforms, there is a problem that I'll need to fix in this version. I'll post an update a bit later today.

 

the signal I posted is periodic, and with version 1.0.50b it gives the best results.

[TomCapraro]

with the previous version the wrong graphs are never generated, this version sometimes generates graphs with a band starting from 1 / 2khz

[pkane2001]

10 minutes ago, TomCapraro said:

the signal I posted is periodic, and with version 1.0.50b it gives the best results.

 

I'll need to fix the "measure simple waveforms" as some changes I introduced to the cross-correlation step are breaking it in this case. Can you please post the best result you get with 1.0.50, so I can make sure 1.0.51 is doing the right thing?

 

[TomCapraro]

Certainly Paul ... !!!

[TomCapraro]

...setting...

[pkane2001]

A quick update to version 1.0.52 fixing some regression items:

• Fix: error on some computers with no permission to query memory statistics
• Fix: filter 1 settings for LP and HP filters @ start were not being processed
• Fix: measure simple waveforms option interrupted alignment process (@TomCapraro - please check)  

Thank you for all the bug reports, and please continue to test!

[TomCapraro]

It seems to work fine.
If I find any anomaly I will report it to you.
Many thanks Paul.

 

👍

[fas42]

Okay, got to get back into the rhythm of playing with DW - and returning to the good ol' Gearslutz loop WAVs, 🙂 ... noting that when setting upsampling to 176400 that different runs with nominally the same other settings are giving me very different results for nulls, and bits reduction for 50% match  - was there always such a variation, depending upon which way the wind was blowing? 😁

 

More concerning, when comparing Original2 with Orion32gen3ch31-32o-ipatchbay, the latter a very recent entry at Gearslutz, I had to discard about half the waveform, to not run out of memory, when upsampled to 176k. Which was fine, but then I tried this,

 

DeltaWave v1.0.51, 2020-10-17T22:22:42.7122571+11:00
Reference:  Original2.wav[L] 2408448 samples 44100Hz 24bits, stereo, MD5=00
Comparison: EchoAudiofire2_Mytek(dheming).wav[L] 2408448 samples 44100Hz 24bits, stereo, 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 - 0Hz, EQ Threshold: -160dB
    Correct Drift:True, Precision:30
    Non-Linear drift Correction:False
    Upsample:True, Window:Hann
    Spectrum Window:Hann, Spectrum Size:32768
    Spectrogram Window:Hann, Spectrogram Size:4096, Spectrogram Steps:2048
    Dither:False
    Trim Silence:False
    Enable Simple Waveform Measurement: False

Resampled Reference to 176400Hz
Resampled Comparison to 176400Hz
Discarding Reference:  Start=0s, End=67s
Discarding Comparison: Start=0s, End=67s

Initial peak values Reference: -0.681dB   Comparison: -2.594dB
Initial RMS values Reference: -17.843dB   Comparison: -19.068dB

Null Depth=13.219dB
X-Correlation offset: -620 samples
Stopped! One or more errors occurred.
Signature: 1b7ef198e0dc8e12a630110d995b0b4e

 

Only by discarding all but 5 secs could I get this to complete ... a memory management issue?

[pkane2001]

4 hours ago, fas42 said:

Okay, got to get back into the rhythm of playing with DW - and returning to the good ol' Gearslutz loop WAVs, 🙂 ... noting that when setting upsampling to 176400 that different runs with nominally the same other settings are giving me very different results for nulls, and bits reduction for 50% match  - was there always such a variation, depending upon which way the wind was blowing? 😁

 

More concerning, when comparing Original2 with Orion32gen3ch31-32o-ipatchbay, the latter a very recent entry at Gearslutz, I had to discard about half the waveform, to not run out of memory, when upsampled to 176k. Which was fine, but then I tried this,

 

DeltaWave v1.0.51, 2020-10-17T22:22:42.7122571+11:00
Reference:  Original2.wav[L] 2408448 samples 44100Hz 24bits, stereo, MD5=00
Comparison: EchoAudiofire2_Mytek(dheming).wav[L] 2408448 samples 44100Hz 24bits, stereo, 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 - 0Hz, EQ Threshold: -160dB
    Correct Drift:True, Precision:30
    Non-Linear drift Correction:False
    Upsample:True, Window:Hann
    Spectrum Window:Hann, Spectrum Size:32768
    Spectrogram Window:Hann, Spectrogram Size:4096, Spectrogram Steps:2048
    Dither:False
    Trim Silence:False
    Enable Simple Waveform Measurement: False

Resampled Reference to 176400Hz
Resampled Comparison to 176400Hz
Discarding Reference:  Start=0s, End=67s
Discarding Comparison: Start=0s, End=67s

Initial peak values Reference: -0.681dB   Comparison: -2.594dB
Initial RMS values Reference: -17.843dB   Comparison: -19.068dB

Null Depth=13.219dB
X-Correlation offset: -620 samples
Stopped! One or more errors occurred.
Signature: 1b7ef198e0dc8e12a630110d995b0b4e

 

Only by discarding all but 5 secs could I get this to complete ... a memory management issue?

 

Hi Frank,

 

What's wrong with just running the test at the correct sampling rate, without introducing additional processing artifacts and increasing memory load? 

 

Just tried Original2 vs Orion32gen3ch31-32o-ipatchbay upsampled to 176k and the result was this (without trimming), but Windows did nearly run out of memory:

• Gain= -0.2057dB (0.9766x) DC=0 Phase offset=0.000063ms (0.011 samples)
• Difference (rms) = -48.81dB [-54.35dBA]
• Correlated Null Depth=60.21dB [53.99dBA]
• Clock drift: 0 ppm
   

From Gearslutz, the result was nearly the same at 44.1k: 399 ns, 0.2046 dB (L), 0.2382 dB (R), -48.8590 dBFS (L), -49.8813 dBFS (R), so I don't see any reason to upsample, but upsampling works... except that it requires lots more memory and processing time ;)

[fas42]

5 hours ago, pkane2001 said:

 

Hi Frank,

 

What's wrong with just running the test at the correct sampling rate, without introducing additional processing artifacts and increasing memory load? 

 

Just tried Original2 vs Orion32gen3ch31-32o-ipatchbay upsampled to 176k and the result was this (without trimming), but Windows did nearly run out of memory:

• Gain= -0.2057dB (0.9766x) DC=0 Phase offset=0.000063ms (0.011 samples)
• Difference (rms) = -48.81dB [-54.35dBA]
• Correlated Null Depth=60.21dB [53.99dBA]
• Clock drift: 0 ppm
   

From Gearslutz, the result was nearly the same at 44.1k: 399 ns, 0.2046 dB (L), 0.2382 dB (R), -48.8590 dBFS (L), -49.8813 dBFS (R), so I don't see any reason to upsample, but upsampling works... except that it requires lots more memory and processing time ;)

 

Mainly, because the matching is vastly better! I got 96dBA for the correlated null depth, when looking at half of the total waveform - so, which is correct, a 53dB null, or a 96dB null? ... Someone wants to know, 😉.

 

Playing the difference waveform of the 96dB result gave fairly random noise, with barely any of the original audible - which implies a subjectively highly accurate loop result. Which is not something that a 53dB number would indicate.

[pkane2001]

47 minutes ago, fas42 said:

 

Mainly, because the matching is vastly better! I got 96dBA for the correlated null depth, when looking at half of the total waveform - so, which is correct, a 53dB null, or a 96dB null? ... Someone wants to know, 😉.

 

Playing the difference waveform of the 96dB result gave fairly random noise, with barely any of the original audible - which implies a subjectively highly accurate loop result. Which is not something that a 53dB number would indicate.

Best to ignore the correlated null number when comparing different sampling rates. It’s related to a product of amplitude and timing errors. and timing errors will be of a completely different scale (per sample) between sampling rates.
 

Correlated null can be useful when comparing the same sampling rates.

[fas42]

Okay, from the perspective of deciding how different two waveforms are, here are two runs - only variation is selecting resampling:

 

DeltaWave v1.0.51, 2020-10-19T09:48:37.1153139+11:00
Reference:  Original2.wav[L] 5340666 samples 44100Hz 24bits, stereo, MD5=00
Comparison: Orion32gen3ch31-32o-ipatchbay.wav[L] 5364167 samples 44100Hz 24bits, stereo, 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 - 0Hz, EQ Threshold: -160dB
    Correct Drift:True, Precision:30
    Non-Linear drift Correction:False
    Upsample:True, Window:Hann
    Spectrum Window:Hann, Spectrum Size:32768
    Spectrogram Window:Hann, Spectrogram Size:4096, Spectrogram Steps:2048
    Dither:False
    Trim Silence:False
    Enable Simple Waveform Measurement: False

Resampled Reference to 176400Hz
Resampled Comparison to 176400Hz
Discarding Reference:  Start=0s, End=0s
Discarding Comparison: Start=0s, End=0s

Initial peak values Reference: -0.526dB   Comparison: -0.636dB
Initial RMS values Reference: -17.35dB   Comparison: -17.681dB

Null Depth=16.342dB
X-Correlation offset: 182 samples
Drift computation quality, #1: Excellent (0.85µs)

Trimmed 0 samples ( 0.00ms) front, 0 samples ( 0.00ms end)

Final peak values Reference: -0.526dB   Comparison: -0.526dB
Final RMS values Reference: -17.35dB   Comparison: -17.35dB

Gain= -0.3121dB (0.9647x) DC=0 Phase offset=1.02943ms (181.592 samples)
Difference (rms) = -81.67dB [-93.02dBA]
Correlated Null Depth=85.6dB [98.08dBA]
Clock drift: 0 ppm

Files are NOT a bit-perfect match (match=10.54%) at 16 bits
Files are NOT a bit-perfect match (match=0.04%) at 24 bits
Files match @ 49.9944% when reduced to 13.38 bits

---- Phase difference (full bandwidth): 81.7608101146944°
    0-10kHz: 0.02°
    0-20kHz: 0.07°
    0-24kHz: 32.57°
---- Variable Group Delay. Frequency matched from 0Hz to 87.2kHz:
    1kHz = 20.4µs (7.33°)
    2kHz = 8.5µs (6.14°)
    4kHz = 2.3µs (3.37°)
    8kHz = 8.8ns (0.03°)
    16kHz = 14.5µs (83.26°)
Timing error (rms jitter): 53.8ns

RMS of the difference of spectra: -158.863578133037dB
gn=1.03659093276496, dc=2.71483673759666E-12, dr=-2.73E-10, of=181.5915002956

DONE!

Signature: b87c0519de0fdfa24659adb0fe19d9be

 

------

 

DeltaWave v1.0.51, 2020-10-19T09:54:23.6030157+11:00
Reference:  Original2.wav[L] 5340666 samples 44100Hz 24bits, stereo, MD5=00
Comparison: Orion32gen3ch31-32o-ipatchbay.wav[L] 5364167 samples 44100Hz 24bits, stereo, 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 - 0Hz, EQ Threshold: -160dB
    Correct Drift:True, Precision:30
    Non-Linear drift Correction:False
    Upsample:False, Window:Hann
    Spectrum Window:Hann, Spectrum Size:32768
    Spectrogram Window:Hann, Spectrogram Size:4096, Spectrogram Steps:2048
    Dither:False
    Trim Silence:False
    Enable Simple Waveform Measurement: False

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

Initial peak values Reference: -0.536dB   Comparison: -0.723dB
Initial RMS values Reference: -17.35dB   Comparison: -17.681dB

Null Depth=17dB
X-Correlation offset: 46 samples
Drift computation quality, #1: Excellent (1.71μs)

Trimmed 0 samples ( 0.00ms) front, 0 samples ( 0.00ms end)

Final peak values Reference: -0.536dB   Comparison: -0.525dB
Final RMS values Reference: -17.35dB   Comparison: -17.35dB

Gain= -0.3106dB (0.9649x) DC=0 Phase offset=1.02961ms (45.406 samples)
Difference (rms) = -51.45dB [-51.73dBA]
Correlated Null Depth=55.8dB [51.22dBA]
Clock drift: -0.04 ppm

Files are NOT a bit-perfect match (match=0.58%) at 16 bits
Files are NOT a bit-perfect match (match=0%) at 24 bits
Files match @ 50.0082% when reduced to 8.51 bits

---- Phase difference (full bandwidth): 5.41687503910554°
    0-10kHz: 2.59°
    0-20kHz: 4.26°
    0-24kHz: 5.42°
---- Variable Group Delay. Frequency matched from 0Hz to 21.1kHz:
    1kHz = 8.9μs (3.19°)
    2kHz = 8μs (5.73°)
    4kHz = 7.8μs (11.25°)
    8kHz = 4.3μs (12.51°)
    16kHz = 10.8μs (62.01°)
Timing error (rms jitter): 1.3μs

RMS of the difference of spectra: -99.2265416896544dB
gn=1.03640886120695, dc=0, dr=-4.327E-08, of=45.4058125879

DONE!

Signature: 288be5bb1347f5d05f86558fa1a9c3c3

 

Needless to say, the difference waveforms from these two runs are completely different,  in every possible way - so, which run is more, 'correct'?

[fas42]

Paul, in the Manual Corrections window I happened to start doing an Optimize run, stopped it, and then did an Apply. The Optimize altered the Gain, but not in the fields at the top; in the Apply I only altered the Offset - but when the results came in, it showed the Offset correctly, but not the Gain Factor specified at the top. So now I don't know which Gain was actually used in that run.

[pkane2001]

3 hours ago, fas42 said:

Okay, from the perspective of deciding how different two waveforms are, here are two runs - only variation is selecting resampling:

 

DeltaWave v1.0.51, 2020-10-19T09:48:37.1153139+11:00
Reference:  Original2.wav[L] 5340666 samples 44100Hz 24bits, stereo, MD5=00
Comparison: Orion32gen3ch31-32o-ipatchbay.wav[L] 5364167 samples 44100Hz 24bits, stereo, 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 - 0Hz, EQ Threshold: -160dB
    Correct Drift:True, Precision:30
    Non-Linear drift Correction:False
    Upsample:True, Window:Hann
    Spectrum Window:Hann, Spectrum Size:32768
    Spectrogram Window:Hann, Spectrogram Size:4096, Spectrogram Steps:2048
    Dither:False
    Trim Silence:False
    Enable Simple Waveform Measurement: False

Resampled Reference to 176400Hz
Resampled Comparison to 176400Hz
Discarding Reference:  Start=0s, End=0s
Discarding Comparison: Start=0s, End=0s

Initial peak values Reference: -0.526dB   Comparison: -0.636dB
Initial RMS values Reference: -17.35dB   Comparison: -17.681dB

Null Depth=16.342dB
X-Correlation offset: 182 samples
Drift computation quality, #1: Excellent (0.85µs)

Trimmed 0 samples ( 0.00ms) front, 0 samples ( 0.00ms end)

Final peak values Reference: -0.526dB   Comparison: -0.526dB
Final RMS values Reference: -17.35dB   Comparison: -17.35dB

Gain= -0.3121dB (0.9647x) DC=0 Phase offset=1.02943ms (181.592 samples)
Difference (rms) = -81.67dB [-93.02dBA]
Correlated Null Depth=85.6dB [98.08dBA]
Clock drift: 0 ppm

Files are NOT a bit-perfect match (match=10.54%) at 16 bits
Files are NOT a bit-perfect match (match=0.04%) at 24 bits
Files match @ 49.9944% when reduced to 13.38 bits

---- Phase difference (full bandwidth): 81.7608101146944°
    0-10kHz: 0.02°
    0-20kHz: 0.07°
    0-24kHz: 32.57°
---- Variable Group Delay. Frequency matched from 0Hz to 87.2kHz:
    1kHz = 20.4µs (7.33°)
    2kHz = 8.5µs (6.14°)
    4kHz = 2.3µs (3.37°)
    8kHz = 8.8ns (0.03°)
    16kHz = 14.5µs (83.26°)
Timing error (rms jitter): 53.8ns

RMS of the difference of spectra: -158.863578133037dB
gn=1.03659093276496, dc=2.71483673759666E-12, dr=-2.73E-10, of=181.5915002956

DONE!

Signature: b87c0519de0fdfa24659adb0fe19d9be

 

------

 

DeltaWave v1.0.51, 2020-10-19T09:54:23.6030157+11:00
Reference:  Original2.wav[L] 5340666 samples 44100Hz 24bits, stereo, MD5=00
Comparison: Orion32gen3ch31-32o-ipatchbay.wav[L] 5364167 samples 44100Hz 24bits, stereo, 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 - 0Hz, EQ Threshold: -160dB
    Correct Drift:True, Precision:30
    Non-Linear drift Correction:False
    Upsample:False, Window:Hann
    Spectrum Window:Hann, Spectrum Size:32768
    Spectrogram Window:Hann, Spectrogram Size:4096, Spectrogram Steps:2048
    Dither:False
    Trim Silence:False
    Enable Simple Waveform Measurement: False

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

Initial peak values Reference: -0.536dB   Comparison: -0.723dB
Initial RMS values Reference: -17.35dB   Comparison: -17.681dB

Null Depth=17dB
X-Correlation offset: 46 samples
Drift computation quality, #1: Excellent (1.71μs)

Trimmed 0 samples ( 0.00ms) front, 0 samples ( 0.00ms end)

Final peak values Reference: -0.536dB   Comparison: -0.525dB
Final RMS values Reference: -17.35dB   Comparison: -17.35dB

Gain= -0.3106dB (0.9649x) DC=0 Phase offset=1.02961ms (45.406 samples)
Difference (rms) = -51.45dB [-51.73dBA]
Correlated Null Depth=55.8dB [51.22dBA]
Clock drift: -0.04 ppm

Files are NOT a bit-perfect match (match=0.58%) at 16 bits
Files are NOT a bit-perfect match (match=0%) at 24 bits
Files match @ 50.0082% when reduced to 8.51 bits

---- Phase difference (full bandwidth): 5.41687503910554°
    0-10kHz: 2.59°
    0-20kHz: 4.26°
    0-24kHz: 5.42°
---- Variable Group Delay. Frequency matched from 0Hz to 21.1kHz:
    1kHz = 8.9μs (3.19°)
    2kHz = 8μs (5.73°)
    4kHz = 7.8μs (11.25°)
    8kHz = 4.3μs (12.51°)
    16kHz = 10.8μs (62.01°)
Timing error (rms jitter): 1.3μs

RMS of the difference of spectra: -99.2265416896544dB
gn=1.03640886120695, dc=0, dr=-4.327E-08, of=45.4058125879

DONE!

Signature: 288be5bb1347f5d05f86558fa1a9c3c3

 

Needless to say, the difference waveforms from these two runs are completely different,  in every possible way - so, which run is more, 'correct'?

 

Ah, now that's a slightly different case since you're using non-linear EQ!  It would be interesting to see if this is true in general or just for certain captures.

 

To answer your question, the result producing lower RMS difference value, lower jitter value, and lower correlated null is 'more correct'. But the reason for it is that DeltaWave can correct for more of the non-linear phase and amplitude effects at a higher resolution and with more averaging to reduce noise when upsampling. 

[pkane2001]

9 minutes ago, fas42 said:

Paul, in the Manual Corrections window I happened to start doing an Optimize run, stopped it, and then did an Apply. The Optimize altered the Gain, but not in the fields at the top; in the Apply I only altered the Offset - but when the results came in, it showed the Offset correctly, but not the Gain Factor specified at the top. So now I don't know which Gain was actually used in that run.

 

Thanks, Frank. Let me check. It sounds like it might be automatically recomputing the gain value even if one is selected manually.

 

[fas42]

As a follow up, only setting the Gain to 0 kicked the processing out of the incorrect behaviour - slight manual alterations of the Gain Factor didn't make this happen.

 

Edit: Spoke too soon! Still hung onto an incorrect Gain in the next run ...

 

[fas42]

When upsampled, and then playing one of the tracks, DW asks if they should be resampled, because my laptop playback can't do that rate. DW seems to hang onto that response, and now immediately plays when requested. But, while playing the upsampled Compare track, and then attempting switching to the Diff track, by clicking the Diff play button, DW got into a loop - and had to be killed. This doesn't happen if the native rate is being used.