Beta-test: DeltaWave Null Comparison software

[pkane]

Yes, correction of the linear part of the total system error, actually the A to B difference. Only small errors, the 1% max regime.
Think applications like capacitor testing, two caps of same nominal value but different construction, say input coupling capacitors.
This will introduce an (additional) highpass in the loopback chain but the corner frequencies will slightly differ (even a after selection and/or analog fine trim) and the resulting phase shift spoils coherence at low frequencies, decreasing null depth. The influence of the linear differences could be brought down one order of magnitude easily if most of it were factored out by applying a well-estimated or measured transfer function. For the capacitor example we know that the correction function would have the shape of an analog (min-phase) shelving filter. In general, any slightly moving pole or zero in the analog transfer function will result in a correction shelving filter (with a fraction of a dB level change).

I'm aware of the problems of processing noise. A way to mitigate this is curve-fitting an analytical transfer function to the reasonably smoothed empirical one, if we know the correction function is min-phase then this can be a bunch of simple IIR filters which can be applied directly (or used to obtain a "clean" convolution kernel from that by sampling a dirac response).

I feel I need to try your program first before making any further comments ;-)

Yes, absolutely. Curve fitting over a large data set is always the way to deal with noise. Choosing the appropriate function to fit is the trick to getting best results. One of the things I've struggled with in DeltaWave is the balance of doing nearly everything automatically without having to keep experimenting with different settings to get the best result. I'm not there yet but the polynomial fit in frequency domain seems to be a reasonable approximation to some of the ADC and DAC filters I have tested so far, although the order of polynomial isn't constant between them. That's making me a bit unhappy at the moment.

 

[pkane]

For your consideration: DeltaWave new version v1.0.4. Some major changes:

• Updated matching algorithm to improve non-linear matching quality
  Linear matching still needs more work. I suggest you stay away from it until further notice
• Added support for ASIO playback
• Remember screen position and size of the main window on restart
• Remember audio driver used in the previous session
• Split the results text into its own tab
• Added some additional menu items to play audio (such as playing Ref and Comp files in left and right channels)

Let me know what you think.

Regards,

-Paul

 

[Alisterkoran]

Hi, I've been reading the forums here for over a year. I've often felt like participating but, well, too many forums, too little time. I would class myself as an 'objective musicphile' (someone once said that an Audiophile uses music to listen to his equipment, a Musicphile uses his equipment to listen to music. YMMV).

Anyway, I found Deltawave intriguing, especially as Diffmaker is useless.

Works very well, so far. Not that I've used it in anger yet, nor explored all the options. I do have a suggestion or two though. On the settings page it's normal practice to have a save button. I know in the instructions you say

Any changes you make to the settings are automatically updated when you close the Setup window
Settings are automatically saved and reloaded the next time you start DeltaWave

but we're all used to having save buttons on settings pages. As it is I keep wondering if the settings really are saved! Also it would be nice to have a 'restore default settings' button on the setup page. As it is, the "Non-Linear Drift Correction" does not stay unchecked, it has to be manually unchecked each time before running a match or the match ends up failing.

Also, I find software a lot easier to use initially if there's a tooltip that shows up when hovering over a setting button, just a short paragraph to give a brief outline of what that setting actually does.

Your website design is excellent, installation is very smooth and trouble free. It would be nice though to have a checkbox during installation asking if a desktop shortcut is wanted or not. But I'm splitting hairs here!

Until now, Diffmaker has been the goto application but sadly it really doesn't work properly and isn't reliable. I hope your Deltawave becomes the reference goto application in future. Thank you for the work you've done creating it and I'm looking forward to following its development.

AK.

 

[Blumlein 88]

Hi, I've been reading the forums here for over a year. I've often felt like participating but, well, too many forums, too little time. I would class myself as an 'objective musicphile' (someone once said that an Audiophile uses music to listen to his equipment, a Musicphile uses his equipment to listen to music. YMMV).

Anyway, I found Deltawave intriguing, especially as Diffmaker is useless.

Works very well, so far. Not that I've used it in anger yet, nor explored all the options. I do have a suggestion or two though. On the settings page it's normal practice to have a save button. I know in the instructions you say
but we're all used to having save buttons on settings pages. As it is I keep wondering if the settings really are saved! Also it would be nice to have a 'restore default settings' button on the setup page. As it is, the "Non-Linear Drift Correction" does not stay unchecked, it has to be manually unchecked each time before running a match or the match ends up failing.

Also, I find software a lot easier to use initially if there's a tooltip that shows up when hovering over a setting button, just a short paragraph to give a brief outline of what that setting actually does.

Your website design is excellent, installation is very smooth and trouble free. It would be nice though to have a checkbox during installation asking if a desktop shortcut is wanted or not. But I'm splitting hairs here!

Until now, Diffmaker has been the goto application but sadly it really doesn't work properly and isn't reliable. I hope your Deltawave becomes the reference goto application in future. Thank you for the work you've done creating it and I'm looking forward to following its development.

AK.

Firstly welcome as this looks like your first post.

I've been testing Deltawave, and it is improving nicely. I tried it with several difficult cases using 1.04 vs Diffmaker. Deltawave in the use cases I've tried returns a better null and of course works. Sometimes the nulls are just a little better, and in some cases more than a little. So I think Paul is on the right track. At various times Diffmaker has gotten a lot of interest from people, and always the sticking point is its tendency to blow up fairly often or now and again give back a result that is obviously wrong. So I hope in time Deltawave will get some widespread attention and use. And thanks to Paul for creating it.

 

[pkane]

Hi, I've been reading the forums here for over a year. I've often felt like participating but, well, too many forums, too little time. I would class myself as an 'objective musicphile' (someone once said that an Audiophile uses music to listen to his equipment, a Musicphile uses his equipment to listen to music. YMMV).

Anyway, I found Deltawave intriguing, especially as Diffmaker is useless.

Works very well, so far. Not that I've used it in anger yet, nor explored all the options. I do have a suggestion or two though. On the settings page it's normal practice to have a save button. I know in the instructions you say
but we're all used to having save buttons on settings pages. As it is I keep wondering if the settings really are saved! Also it would be nice to have a 'restore default settings' button on the setup page. As it is, the "Non-Linear Drift Correction" does not stay unchecked, it has to be manually unchecked each time before running a match or the match ends up failing.

Also, I find software a lot easier to use initially if there's a tooltip that shows up when hovering over a setting button, just a short paragraph to give a brief outline of what that setting actually does.

Your website design is excellent, installation is very smooth and trouble free. It would be nice though to have a checkbox during installation asking if a desktop shortcut is wanted or not. But I'm splitting hairs here!

Until now, Diffmaker has been the goto application but sadly it really doesn't work properly and isn't reliable. I hope your Deltawave becomes the reference goto application in future. Thank you for the work you've done creating it and I'm looking forward to following its development.

AK.

Hi AK,

Thank you for your comments! DeltaWave is work in progress, and as @Blumlein 88 is well aware, I'm open to all suggestions I consider it to be an experimenter's toolbox, and so keep adding more features with every update. Usability and help/tooltips are always lagging, unfortunately.

I'm hoping to get v1.0.5 out with some enhancements and new features very soon, and then pause for an overall clean-up and UI redesign... but only if I don't think of some new cool feature to add, in the meantime

Keep those suggestions coming!

Regards,

-Paul

 

[ofrappier]

Hi,

It's works nicely. Any OSX version ?

I've used OSX with BIT perfect CORE AUDIO, and Windows 10 to compare the original and recorded .wav.


Here a test with a DJ Software routed with Audio Hijack on OSX Mojave, at 16/44.1 like the resolution of the original audio test file :

VIRTUAL DJ 8.3 Software, with AUTO-GAIN off, Pitch quality at 3 in settings, and Main Volume to 0dB :

The results, even with the recorded file recorded with Audio Hijack on OSX :

so, on OSX , AUDIO HIJACK and VIRTUALDJ 8.3 are BIT PERFECT.

Another DJ Software Rekordbox 5.4.3 export mode : With same AUDIO file, Volume at 0db, and same test, without effects :

The second software, isn't bit perfect , with phase problem and Clock Drift problem at 7.94 ppm

So, your software work nicely

Thanks !

P.S :
A precision :
These tests were done with OSX and core audio bit perfect; and the capture software (https://rogueamoeba.com/audiohijack) also perfect bit to record a wav (After check with Delta Wave) Then the samples (original and recorded with audiohijack) are automatically aligned in the analysis software Delta wave in a window of 20 or 60 sec to avoid bit shifts between the beginning and the end of the sample between the two lossless .wav at same 16 / 44.1 resolution So OSX for the comparative records and WIN10 for the comparison software Delta Wave -

Otherwise Replay music 8 on Windows 10, with direct APPS recording feature, is another solution because tested BIT perfect with Delta Wave. On the other hand other software tests with "Wasapi loopback" like Audacity, are not perfect BIT. (after analysis with Delta Wav ).

But the best for the perfect BIT in Shared mode and exclusive mode is the MacOSX Core Audio with audiohijack (https://rogueamoeba.com/audiohijack). More flexible and less complicated than Wasapi / ASIO ....... Even in Exclusive mode, and without bit/res changes, several Windows Virtual cable aren't bit perfect.

 

[pkane]

Hi,

It's works nicely. Any OSX version ?

I've used OSX with BIT perfect CORE AUDIO, and Windows 10 to compare the original and recorded .wav.


Here a test with a DJ Software routed with Audio Hijack on OSX Mojave, at 16/44.1 like the resolution of the original audio test file :

VIRTUAL DJ 8.3 Software, with AUTO-GAIN off, Pitch quality at 3 in settings, and Main Volume to 0dB :
View attachment 22306


The results, even with the recorded file recorded with Audio Hijack on OSX :

View attachment 22307

so, on OSX , AUDIO HIJACK and VIRTUALDJ 8.3 are BIT PERFECT.

Another DJ Software Rekordbox 5.4.3 export mode : With same AUDIO file, Volume at 0db, and same test, without effects :

View attachment 22308

View attachment 22309

The second software, isn't bit perfect , with phase problem and Clock Drift problem at 7.94 ppm

So, your software work nicely

Thanks !

Thank you for testing and reporting! I've had some requests for OSX and Linux versions. It's possible, but will take some work. Let me go through some of the things I still want to add/improve, and then I'll see what I can do

Regards,

-Paul

 

[andreasmaaan]

Great work @pkane! Gonna have a play with this over the weekend

 

[Pluto]

Sorry to rain on your parade, but having just installed the software on my Windows 3840 × 2160 display, everything is far too small. Selecting "disable display scaling on high DPI settings" helps, but experience suggests that this route is not a universal panacea.

Using the text size adjustment control (top right) causes many of the captions (particularly the trim adjustment fields) to overrun the available space and has no effect on the main menu items, which remain tiny.

So this is probably a bridge you will have to cross sooner or later. I can do screen shots if this will help.

 

[pkane]

Sorry to rain on your parade, but having just installed the software on my Windows 3840 × 2160 display, everything is far too small. Selecting "disable display scaling on high DPI settings" helps, but experience suggests that this route is not a universal panacea.

So this is probably a bridge you will have to cross sooner or later.

Try clicking the Aa button at the top right:

 

[Pluto]

Try clicking the Aa button at the top right

As I said when I edited the post (which I think crossed over your post), it didn't help greatly...


Likewise, the tab labels remain almost the same (far too small) and the Windows menu items look as thought they are 6pt (if they were on paper).


The combination of your ‘Aa’ button and “disable display scaling on high DPI settings” makes the software workable, but things tend to get a bit unpredictable when the latter setting has to be used.

Anyway, with an hour of testing the software it didn't crash at all. Well done!

I think the 'busy' spinner needs to be more prominent.

The scale on the various graphs needs to be rounded and abbreviated in a common-sense way, IMHO. I suggest that seconds (to one decimal place) might be more useful than milliseconds. In the same vein, I'm not sure that offering to show the sample number on the x-axis is a great deal of use.

I'm sure that you are fully aware, but the zoom controls really need to be mouse-wheel-driven. Something along the lines of using the wheel to zoom the x-axis, and alt+wheel for the y-axis.

Congratulations - this has the potential to be a fantastic piece of work!

 

[pkane]

As I said when I edited the post (which I think crossed over your post), it didn't help greatly...

View attachment 22313
Likewise, the tab labels remain almost the same (far too small) and the Windows menu items look as thought they are 6pt (if they were on paper).

View attachment 22314
The combination of your ‘Aa’ button and “disable display scaling on high DPI settings” makes the software workable, but things tend to get a bit unpredictable when the latter setting has to be used.

Anyway, with an hour of testing the software it didn't crash at all. Well done!

I think the 'busy' spinner needs to be more prominent.

The scale on the various graphs needs to be rounded and abbreviated in a common-sense way, IMHO. I suggest that seconds (to one decimal place) might be more useful than milliseconds. In the same vein, I'm not sure that offering to show the sample number on the x-axis is a great deal of use.

I'm sure that you are fully aware, but the zoom controls really need to be mouse-wheel-driven. Something along the lines of using the wheel to zoom the x-axis, and alt+wheel for the y-axis.

Congratulations - this has the potential to be a fantastic piece of work!

Thanks for feedback! I think the display issue is more with DPI settings than anything else. I'll see what I can do about this, but more often than not, Windows just doesn't do a good job of managing font sizes and non-standard DPI settings. Here's what the large-font version looks like with the 'stock' Windows DPI and font settings:



Zoom controls should work with the mouse-wheel. Just have the mouse positioned inside the plot while turning it if you want both axis to zoom-in/out, or over one of the axis if you want to zoom just on that axis. You may need to click once anywhere inside the plot to make sure the mouse controls apply to it.

Regards,

-Paul

 

[Pluto]

Zoom controls should work with the mouse-wheel. Just have the mouse positioned inside the plot while turning it if you want both axis to zoom-in/out, or over one of the axis if you want to zoom just on that axis. You may need to click once anywhere inside the plot to make sure the mouse controls apply to it.

OK – that does work (thinks: where exactly do I have to place the pointer to obtain x-axis-only zooming?) but may I suggest an alternative approach?

• That the wheel always drives the x-axis zoom wherever the pointer is within DeltaWave's window. After all, the wheel has no other function while your program has focus
• Likewise, alt-wheel drives the y-axis zoom
• Rather than constantly attempting to rebuild the graph in real-time while the user tweaks the wheel (a hopeless task, and I use a reasonably fast machine), try to provide some kind of indication of what the zoom settings will be doing (a dotted marquee, perhaps), but don't attempt to actually do it until you are reasonably sure that the user has finished tweaking the wheel, then get on with it. I hope I have described that with sufficient clarity to make sense!

 

[pkane]

OK – that does work (thinks: where exactly do I have to place the pointer to obtain x-axis-only zooming?) but may I suggest an alternative approach?

• That the wheel always drives the x-axis zoom wherever the pointer is within DeltaWave's window. After all, the wheel has no other function while your program has focus
• Likewise, alt-wheel drives the y-axis zoom
• Rather than constantly attempting to rebuild the graph in real-time while the user tweaks the wheel (a hopeless task, and I use a reasonably fast machine), try to provide some kind of indication of what the zoom settings will be doing (a dotted marquee, perhaps), but don't attempt to actually do it until you are reasonably sure that the user has finished tweaking the wheel, then get on with it. I hope I have described that with sufficient clarity to make sense!

Thanks, @Pluto, all makes sense! I admit, I spent a long time on the charts in DeltaWave. The open source charting package I used had a lot of good functionality but was never designed to perform well on many millions of data points in double floating format. I had to rewrite portions of the rendering library just to get the plots and the zoom/scroll to become more real-time, more responsive. When I started, a short hi-res track could take minutes to render. It's a bit better now

Here are some things that may not be that obvious about how the zoom works today:

• Where you position the mouse when zooming does matter: the zoom will occur around the point where the mouse is, keeping that point on the screen. This lets you zoom in or out while focusing on a specific artifact or a point of interest. This also works when zooming in around a point on the X or Y axis
• You can always reset the axis back to the original zoom setting by clicking Reset Axis button
• The
  
  icon brings up a dialog to allow you to enter custom left/right/top/bottom dimensions to set the desired zoom level. Units in this window still remain to be worked out, so it's not very intuitive. But, sometimes it's a lot easier to use this when you know the range of where you want to look rather than keep rolling the mouse wheel for that extreme zoom-in
• Similar charts (for example, all spectrum charts, or all waveform charts) remember their zoom level from other similar charts. When you zoom-in on one FFT chart, all the others will show at the same zoom level. This allows you to quickly flip between them for an apples-to-apples comparison.
• The zoom setting you select for your charts on the display will also be used in the generated HTML report, so pick a level that works best for you and then it will be reflected in the report.

Regards,

-Paul

 

[Pluto]

Where you position the mouse when zooming does matter

Indeed – so is it possible, at present, to do X-axis-only zooming? It's quite a common requirement to examine a particular point in time at the full range of level.

The open source charting package I used had a lot of good functionality but was never designed to perform well on many millions of data points in double floating format

It is impressive. I think the trick may be, in due time, for you to simplify what needs to be plotted for a particular graph. To explain that idea a bit further, you are (probably) including in each plot thousands of data points that don't mean very much in the context of, say, a five minute X-axis. Hence, the trick is to only plot those points that have visual significance for the plot in hand. In other words, it's a kind of perceptual encoding – only plot the data that have visual significance!

Also, would there be mileage in rounding out of double float format into integer? 65000-ish data points on the Y axis is more than adequate.

What do you think of my suggestion to delay all plotting until you believe the user has stopped tweaking the wheel?

 

[Pluto]

How much hassle would it be to add DFF as a supported DSD format? It's quite common.

You do realize that the “hate PCM” / “love DSD” brigade will completely dismiss any results produced in this software because of the need to convert to PCM for the calculations. It will be argued that the magic of DSD is lost in the conversion

Is it possible to place a cursor at a point on a graph and 'play from here'? I notice that the playback cursor appears to bleed through to other applications on the screen.

 

[pkane]

Indeed – so is it possible, at present, to do X-axis-only zooming? It's quite a common requirement to examine a particular point in time at the full range of level.

Yes, absolutely. Position the mouse anywhere below the actual X axis, right around where the labels are, click there, and then use the mouse wheel to zoom. It will zoom-in or out at the point where the mouse is, keeping that point on screen, but only zooming on the X-axis. The same works for Y-axis.

Also, would there be mileage in rounding out of double float format into integer? 65000-ish data points on the Y axis is more than adequate.

Yes. That's a bit of a struggle with this plot library, as it picks arbitrary divisions to display tick points. I really want to improve it, but also need the ability to display fractions as one zooms-in, since integer display isn't sufficient at higher zoom levels. Definitely on my to-do list.

What do you think of my suggestion to delay all plotting until you believe the user has stopped tweaking the wheel?

I do like the real-time feedback of the zoom feature as it is now, but can see the value of making it faster, especially for larger charts and slower PCs.

How much hassle would it be to add DFF as a supported DSD format? It's quite common.

If I can find the file format description, I can add it

You do realize that the “hate PCM” / “love DSD” brigade will completely dismiss any results produced in this software because of the need to convert to PCM for the calculations. It will be argued that the magic of DSD is lost in the conversion

Sure! I added DSF support primarily for my own experimenting. I was curious to see if there were any obvious differences with PCM, but found very little.

Is it possible to place a cursor at a point on a graph and 'play from here'? I notice that the playback cursor appears to bleed through to other applications on the screen.

Yes! I'm actually working on adding an ABX-style comparator that lets you listen and compare the waveforms once they are suitably matched in level and phase. Stay tuned.

 

[ofrappier]

Hi,

I've play this week-end with your great software with OSX and WIN10 DJ software or player :

Results here :

Hi,

More tests : (with new https://deltaw.org/, software that compare 2 wav with auto sample alignement)

Original file : 60s - 16/44.1

OSX :

Core Audio OSX : BIT PERFECT
Audio HI JACK : BIT PERFECT in 32 bits or 16 bits recording (https://rogueamoeba.com/audiohijack/)
Loopback virtual cable : BIT PERFECT (https://rogueamoeba.com/loopback/)

System player audio : Bit perfect
Virtual DJ 8.3 OSX (Without effect - gain 0dB - autogain off) + AUDIO HI JACK recording :
100% BIT PERFECT no jitter

Virtual DJ 8.3 OSX (Without effect - gain 0dB - autogain off) + Loopback virtualcable + Audacity :
100 % BIT PERFECT no jitter / with Audacity 32/24 bits recording of 16bits original file.

Rekordbox 5.4.3 OSX (Without effect - gain 0 dB - autogain off) + AUDIO HI JACK recording or Loopback virtualcable + Audacity :
No BIT perfect / Clock jitter problem/ phase- problem/volume change problem

Serato lite 1.1 (Without effect - gain 0 dB - autogain off) + AUDIO HI JACK recording or Loopback virtualcable + Audacity :
No BIT perfect / Clock jitter problem/ phase- problem/volume change problem

Traktor 3.1 OSX (Without effect - gain 0 dB - autogain off) + AUDIO HI JACK recording or Loopback virtualcable + Audacity :
No BIT perfect / Clock jitter problem/ phase problem/volume change problem

Engine 1.2.3 OSX (Without effect - gain 0 dB - autogain off) + AUDIO HI JACK recording or Loopback virtualcable + Audacity :
No BIT perfect / phase problem/volume change problem - clock ok

DJAY PRO 2 OSX (Without effect - gain 0 dB - autogain off) + AUDIO HI JACK recording or Loopback virtualcable + Audacity :
No BIT perfect / phase problem/volume change problem -

WIN 10 :

WIN10 : BIT PERFECT with Wasapi
Voicemeter Virtual ASIO or Wasapi : 99% BIT PERFECT with Voicemeter Banana recording in 32/24 bits of a 16 bits Files. The 16 bits recording isn't bit perfect
(https://www.vb-audio.com/Voicemeeter/banana.htm)

Windows Wasapi loopback with audacity : not perfect - gain and phase problems (The windows Wasapi loopback isn't BIT perfect)

Foobar WIN10 (Without effect - gain 0dB - no replay gain ) + Voicemeter Bana recording virtual ASIO : 99.8 BIT PERFECT no jitter

Virtual DJ 8.3 WIN10 (Without effect - gain 0dB - autogain off) + Voicemeter Bana recording virtual ASIO : 99.8 BIT PERFECT no jitter
Virtual DJ 8.3 WIN10 (Without effect - gain 0dB - autogain off) + Voicemeter virtual asio or Virtual Wasapi + Audacity recording :
99.8% BIT PERFECT no jitter / with Audacity 32/24 bits recording of 16bits original file.

Rekordbox 5.4.3 Win 10 (Without effect - gain 0 dB - autogain off) internal recording at any levels or with Voicemeter virtual asio or Virtual Wasapi + Audacity recording Audacity 32/24 bits:
No BIT perfect / Clock jitter problem/ phase- problem/volume change problem

Serato lite 1.1 Win 10 (Without effect - gain 0 dB - autogain off) internal recording at any levels or with Voicemeter virtual asio or Virtual Wasapi + Audacity recording Audacity 32/24 bits:
No BIT perfect / Clock jitter problem/ phase- problem/volume change problem

Traktor 3.1 Win 10 (Without effect - gain 0 dB - autogain off) internal recording at any levels or with Voicemeter virtual asio or Virtual Wasapi + Audacity recording Audacity 32/24 bits:
No BIT perfect / Clock jitter problem/ phase problem/volume change problem

Engine 1.2.2 Win 10 (Without effect - gain 0 dB - autogain off) i with Voicemeter virtual asio or Virtual Wasapi + Audacity 32/24 bits recording :
No BIT perfect / phase problem/volume change problem - clock ok

DJAY PRO WIN10 (Without effect - gain 0 dB - autogain off) Voicemeter virtual asio or Virtual Wasapi + Audacity 32/24 bits recording :
No BIT perfect / phase problem/volume change problem

Great Software ! i'm not blind know

 

[pkane]

Great Software ! i'm not blind know

Hi! And thank you for testing!

The blind part is coming soon, too Here's a preview of the ABX comparator:

 

[Rock Rabbit]

Testing 1.05
I have problems comparing dsd 64 vs flac 24/192 (1 min, same track, no dither), residual drift error too large.
Continue anyway shows a big step on matched spectra, if the reference is DSF the step is on audio range. But if FLAC is reference the step is on ultrasonic range.
The spectra matches well on audio but it will never over 30 kHz due to DSD nature...
WASAPI or ASIO only playing at 44.1