Beta-test: DeltaWave Null Comparison software

[Rantapossu]

The initial missing config file error will be fixed, along with the missing reference file, if none are selected.
WASAPI exclusive is the preferred way to measure, which is why it's the default.
I'll think about turning on auto-trim. I do agree it's useful in many more cases than it's not

Great!

Have you thought about replacing the older "Compare" (and "Delta") -file only after the successful recording and processing of the new one and not when opening the recorder window from the menu?

 

[Grooved]

Hi @pkane

In the record mode, when I try both ASIO with Tone2 Pro and Cosmos ADC, I get this error message:
"Can't play audio (Khadas Tone 4.1kHz)
invalid number of input channels 2, must be in the range of [0,0]"

Why it is expecting anything on input channels from the device selected as output?

 

[pkane]

Hi @pkane

In the record mode, when I try both ASIO with Tone2 Pro and Cosmos ADC, I get this error message:
"Can't play audio (Khadas Tone 4.1kHz)
invalid number of input channels 2, must be in the range of [0,0]"

Why it is expecting anything on input channels from the device selected as output?

That’s due to the way the ASIO driver initialize call works in the audio library I’m using. It tries to open both, input and output channels in one call. I fixed this is Multitone so that it works correctly, but that fix is not yet in DeltaWave.

 

[JJ_89]

Hello all,
I'm new to DeltaWave and audio testing as a whole. Someone recommended that I use this software and I'm somewhat confused as to what the different measurements are. Could someone explain which of the result metrics are most important and what they mean? I understand what the graphs for waveforms and spectra are, however, I get confused at the results given in the generated reports. Is there a guide to these metrics here or somewhere else online?

P.S. I’m comparing a reference WAV to different quality MP3 LAME files.

 

[staticV3]

Is there a guide to these metrics here or somewhere else online?

you can find a brief description of the different graphs here: https://deltaw.org/mydoc_quickstart.html

 

[JJ_89]

you can find a brief description of the different graphs here: https://deltaw.org/mydoc_quickstart.html

Thanks alot. I am genuinely shocked that I was too dumb to check the website that I downloaded the app from for help. Is there any guide to the more complex values that can be found in the generated reports?

 

[pkane]

Version 2.0.8 of DeltaWave is now available with some fixes and enhancements, primarily related to the Loop Recorder feature.

Changes in 2.0.8​

• Fixed: Loopback recorder failed when using different ASIO drivers for input/output
• Fixed: Recorder not restoring settings when re-invoked
• Fixed: Error after the first installation because the default settings file was not found
• Fixed: One empty channel recorded in ‘stereo’ mode with ASIO
• Fixed: Recorder can now be launched with or without reference/comparison files specified in the main window
• Fixed: Error “Invalid number of input channels 2” when using ASIO output device
• Fixed: recorder always plays both channels, even if only one channel is selected
• Changed: removed option to launch DeltaWave from the installer – was causing errors due to some installation tasks still being incomplete

@Rantapossu, @Grooved -- please be gentle, don't break too many things at once!

 

[Rantapossu]

@Rantapossu, @Grooved -- please be gentle, don't break too many things at once!

Thanks!

Superb job! Almost all of my previous findings are now fixed. There's still one old bug left, but its behaviour has changed slightly.

How to replicate:

1. Make a new recording and wait until the results are ready.
2. Open the "Record" -window again, but don't press the Record-button.
3. Close the "Record" -window.
4. The "Compare" -file changes now incorrectly by itself, it's now a bit-perfect copy of the "Reference" -file. This shouldn't happen if you cancel your recording before pressing the Record-button. It should only change after a successful recording and even after that it should be real result and not a direct copy of the original.
5. Because the "Compare" -file and the "Reference" -file are now identical, the "Match" -file changes itself as a digital silence too.

I think that even starting the record and pressing the "Stop" button during the record should not replace the "Compare" -file, it should replace it only when the recording is ready and matching begins.

 

[Rantapossu]

This is more like a beauty flaw than a bug, but during the new recording the results of the previous recording are incorrectly visible on the bottom of the "Record" -window:

(The Fs in and Fs out might be from the previous recording as well, I didn't check it. The "Level" and "Difference" are for sure.)

The correct values become visible after the matching is ready and the old values are overwritten.

Could the results be blank during recording and matching (as they are during the first recording after the fresh start of DW), because you can see the previous values on the bottom of the main window?

 

[pkane]

This is more like a beauty flaw than a bug, but during the new recording the results of the previous recording are incorrectly visible on the bottom of the "Record" -window:

View attachment 247974

(The Fs in and Fs out might be from the previous recording as well, I didn't check it. The "Level" and "Difference" are for sure.)

The correct values become visible after the matching is ready and the old values are overwritten.

Could the results be blank during recording and matching (as they are during the first recording after the fresh start of DW), because you can see the previous values on the bottom of the main window?

This one I'm reluctant to change, as I often use this as a way to compare to the previous result. If the number gets cleared when I start recording, I'll likely forget it before the new one shows up I'll probably change the color or give some other indication that the values are stale...

Thanks for testing and reporting these issues!

 

[Rantapossu]

This one I'm reluctant to change, as I often use this as a way to compare to the previous result. If the number gets cleared when I start recording, I'll likely forget it before the new one shows up I'll probably change the color or give some other indication that the values are stale...

Thanks for testing and reporting these issues!

No problem!

How about two rows one on top of the other, the previous recording (visible all the time during the current run) and the new recording (blank until the matching is ready)?

There could be even the third row showing the offset between the new and old recording...

 

[Grooved]

Hi, so after a quick test:

Changes in 2.0.8

• Fixed: Loopback recorder failed when using different ASIO drivers for input/output

Not sure if it was not working at all or with a few devices only, but it looks OK with the two I just tried

• Fixed: Recorder not restoring settings when re-invoked

I launch the first test, I tried both at 44.1, then 44.1 out and 176.4 in, then back to both 44.1
I didn't change anything from the first test, so it should have use the same reference file, and I should have got the same result in the third test than in the first one, which was not the case as I got this message in the third test

• Fixed: Error after the first installation because the default settings file was not found

Looks OK

• Fixed: Recorder can now be launched with or without reference/comparison files specified in the main window

Looks OK too

• Fixed: Error “Invalid number of input channels 2” when using ASIO output device

This one is fixed for sure now

Not perfect for now, but another huge thanks for the hard work @pkane

 

[pkane]

Hi, so after a quick test:


Not sure if it was not working at all or with a few devices only, but it looks OK with the two I just tried


I launch the first test, I tried both at 44.1, then 44.1 out and 176.4 in, then back to both 44.1
I didn't change anything from the first test, so it should have use the same reference file, and I should have got the same result in the third test than in the first one, which was not the case as I got this message in the third test
View attachment 248066


Looks OK


Looks OK too


This one is fixed for sure now

Not perfect for now, but another huge thanks for the hard work @pkane

Getting there Different sampling rates at output and input may require some different settings (upsample or downsample to the desired rate, trim one or both files, etc.) in settings. I suggest downsample to 44.1k to see if that works any better.

 

[Grooved]

Getting there Different sampling rates at output and input may require some different settings (upsample or downsample to the desired rate, trim one or both files, etc.) in settings. I suggest downsample to 44.1k to see if that works any better.

I will look at it, but the problem is that my third test was back to 44.1-44.1, like the first one, so I supposed that DeltaWave did not use the same reference track than in the first test, or I would have got the same result in first and third tests

 

[slartibart]

If I make multiple recordings in Audacity of my DAC analog output through a Focusrite Scarlett 2i2 the results can be fairly variable when compared with Deltawave. At one extreme the clock drift doesn't converge and I am asked if I want to try an alternative method which never gives a good result and appears to invert one of the waveforms. At the other extreme the really good results have much less phase change and group delay at higher frequencies compared to the poor results.
Is it possible that the alignment of the waveforms to subsample accuracy doesn't always work as well as it does for the best results?

 

[pkane]

If I make multiple recordings in Audacity of my DAC analog output through a Focusrite Scarlett 2i2 the results can be fairly variable when compared with Deltawave. At one extreme the clock drift doesn't converge and I am asked if I want to try an alternative method which never gives a good result and appears to invert one of the waveforms. At the other extreme the really good results have much less phase change and group delay at higher frequencies compared to the poor results.
Is it possible that the alignment of the waveforms to subsample accuracy doesn't always work as well as it does for the best results?

Can't really tell what's going on without some more details as to what you're doing, what's on the recording, what settings you're using, and sharing a recording or two that others could look at and analyze.

 

[slartibart]

Can't really tell what's going on without some more details as to what you're doing, what's on the recording, what settings you're using, and sharing a recording or two that others could look at and analyze.

OK here are three example files. 100.wav is the reference. my Deltawave thinks mpd3 is reversed polarity but I told it not to invert. I see more than 10dB difference in the spectrum of delta. I deleted all the really bad ones but I can capture some more later.
https://www.dropbox.com/s/53spkr9hglakrl3/Dropbox.zip?dl=0

 

[pma]

Hello Paul, I have come to an interesting problem today. Normally I prepare my files for listening tests through the soundcard which has the same clock for its ADC and DAC and when comparing the loop product to the original music file, Deltawave gives negligibly low error numbers and the files are indistinguishable in the ABX test. Today, I have decided the loop constituted of Topping D10s DAC (connected to PC through USB-ISO to cut the ground loop) and E1DA Cosmos ADC. For THD or IMD measurements, this loop gives excellent numbers. However, in a loop with a music file, and the result compared to the original file (both previously level matched and time aligned in AA), the result is horrible. Not only in pkmetrics and error distribution, but the difference between the original and the loop product is easily audible. Would you have an idea what could be the issue? It is definitely not the Deltawave itself.

 

[KSTR]

@pma, could you post the files (snippet of 30seconds or so)?

IME, non-synced record-while-playback is fraught with problems for subtractive analysis as DW cannot correct dynamic clock drift, only average clock drift. And then you have non-corrected magnitude and phase errors from the DAC and ADC (roll-offs and passband ripple), therefore I usually use steep filters (pre and post, and on ref and test) in DW to avoid large residuals only comming from roll-offs. If possible, choose linear phase filters in DAC and ADC. As a last resort, use "nonlinear calibration". Another trick I'm using is recording the transfer function of the loopback and apply the obtained IR on the original file before compare, this eliminates any mag/phase differences and gives much better null and less artifacts in the residual than Paul's "non-linear calibration". But that requires sample-synced record-while-playback to be fully successful.
The lack of syncing means is what kept me from using a Cosmos ADC so far.

Owning a D10 balanced, I found that this DAC has a lot of dynamic clock drift (unstable PLL) and I would guess the D10s is similar as it will likely operate in async mode internal (the final DAC core running off of a 100MHz crystal, resampling everything asynchronously). I don't know if the ADC is async mode as well...
And who knows, maybe you are experiencing the infamous ESS sound ;-)

 

[pma]

@pma, could you post the files (snippet of 30seconds or so)?

Sure. Here are the files

https://pmacura.cz/bathroomwin.zip

Below is my ABX result (repeatedly)

foo_abx 2.0.2 report
foobar2000 v1.4.8
2022-12-30 20:48:50

File A: Track_13_orig.wav
SHA1: 4699325addcdb094389a092292d93f17449995cc
File B: Track13_Loopcut16.wav
SHA1: 924125c802abcff2818cbf25db96a05829a90bd0

Output:
ASIO : Topping USB Audio Device
Crossfading: NO

20:48:50 : Test started.
20:51:35 : 01/01
20:51:51 : 02/02
20:52:08 : 03/03
20:52:17 : 04/04
20:52:27 : 05/05
20:52:38 : 06/06
20:52:47 : 07/07
20:52:56 : 08/08
20:53:07 : 09/09
20:53:20 : 10/10
20:53:30 : 11/11
20:53:38 : 12/12
20:53:48 : 13/13
20:53:57 : 14/14
20:54:07 : 15/15
20:54:17 : 16/16
20:54:17 : Test finished.

 ----------
Total: 16/16
Probability that you were guessing: 0.0%

 -- signature --
72b0b470a3d945575cf88e5ad89db0f7fd20a53b

I concentrate at the guitar riff in the left ear [ she came in through the bathroom window ... - now the riff, in the original is sharp, in the loop is muddy, lifeless ]