Beta Test: Multitone Loopback Analyzer software

[pkane]

I don't see that in the V5.20.9 release or later builds, which version are you using Paul?

Hey John,

This may just be an issue with FlexASIO.. Just updated to 5.20.9 and still get the same artifacts (REW generator and REW RTA capture over FlexASIO):

 

[KSTR]

REW played using FlexASIO driver into digital loopback driver. Multitone was recording over WASAPi, same driver.

Ah OK, with indirection ... where many things can happen...

 

[KSTR]

NP. I've been back and forth on conversions lately trying (and failing) to find a universal solution, so for the next build I have added a check box to control whether 32-bit data should be treated as 24-bit or not since there are some situations where resolution can be higher than 24-bit. Progress is also being made (with Pavel Hofman's generous help) on direct WASAPI exclusive support

I for one never had an issue with REW under windows using native ASIO drivers.
I do have the small artifact issue with REW under Linux (as posted here), so is it likely the new 24/32-setting would fix that also?

 

[pkane]

Ah OK, with indirection ... where many things can happen...

Are you implying that the Windows audio stack is in any way a "direct" way to play audio? What a mess...

 

[KSTR]

Are you implying that the Windows audio stack is in any way a "direct" way to play audio? What a mess...

I'm not implying it. But I quickly see when it's not a bit-transparent setup, RME bit-test wav's are a true life-saver.
ASIO never failed me, 24bit pass with almost all player applications. REAPER even manages to clear the 32bit test.

 

[JohnPM]

Hey John,

This may just be an issue with FlexASIO.. Just updated to 5.20.9 and still get the same artifacts (REW generator and REW RTA capture over FlexASIO):

What's the sample rate? Clean for me with FlexASIO or RME's Madiface ASIO driver.

 

[KSTR]

To add to my last post... that's why I would like to test the stereo file playback in MT as soon as posible, to see if bittest passes.

 

[pkane]

I'm not implying it. But I quickly see when it's not a bit-transparent setup, RME bit-test wav's are a true life-saver.
ASIO never failed me, 24bit pass with almost all player applications. REAPER even manages to clear the 32bit test.

DeltWave does the same bit-perfect test but with any loopback device(s) and any waveform. I can add this test to MT, as well.

 

[KSTR]

I can add this test to MT, as well.

Not really required from my side...
Any comments on what might be going on with my post #830 problem, which is really bizarre, isn't it?

 

[pkane]

Not really required from my side...
Any comments on what might be going on with my post #830 problem, which is really bizarre, isn't it?

If you're talking about the improved low-level performance at -0.001 gain level, I wonder if this has to do with CAudioLimiter that @edechamps pointed out. As I understand it, shouldn't be an issue with WASAPI Exclusive mode, but... that's what it suspiciously looks like.

 

[pkane]

What's the sample rate? Clean for me with FlexASIO or RME's Madiface ASIO driver.

44.1k, but not using RME or Madiface driver. I'll need to test with RME loopback on a different computer. Always makes testing difficult when conditions are different due to switching drivers, devices and computers I also use different versions of Windows between these, so that doesn't help to get consistent results!

 

[KSTR]

If you're talking about the improved low-level performance at -0.001 gain level, I wonder if this has to do with CAudioLimiter that @edechamps pointed out. As I understand it, shouldn't be an issue with WASAPI Exclusive mode, but... that's what it suspiciously looks like.

Or any other gain like +0.1dB or whatever, provided it doesn't clip the signal.
To recap:
- I exported REW or MT's own 50Hz signal at -1dB into a 32float file. Checked the file with REW's file-RTA, clean.
- Use that as test signal file, import it.
- With 0dB playback gain (to use the level in the file as is) I get artifacts even with WASAPI.
- Changing the gain gives clean result, but only with WASAPI.
- Using internal 50Hz/-1dB again and ASIO driver, I get pretty much the exact same error pattern than with the 0dB WASAPI file playback (only playback driver seems to play a role, btw). So I would assume it is the same or very close error mechanism at work.

 

[pkane]

To recap:
- I exported REW or MT's own 50Hz signal at -1dB into a 32float file. Checked the file with REW's file-RTA, clean.

Seems strange. It would make it easier to troubleshoot if I could get a clean result with REW, but for some reason I don't, and in my testing, both REW and MT produce nearly identical results as I posted earlier.

 

[KSTR]

Seems strange. It would make it easier to troubleshoot if I could get a clean result with REW, but for some reason I don't, and in my testing, both REW and MT produce nearly identical results as I posted earlier.

But you do get a clean result of a MT-generated 32float-file when dropped into REW's RTA window, don't you? It is basically impossible to get a different result (other than a somehow broken version of REW).
This proves the file itself is OK (as is another one created by REW, this time). And there aren't any soundcards involved yet.

Now, using that file in MT for actual loopback testing, it tests artifact-free, provided that the output driver is WASAPI/ex and the play gain is not exactly 0dB.

- the file itself is clean
- load and direct play as is in MT yet shows errors even with WASAPI.
- changing the gain "fixes" it.

 

[pkane]

But you do get a clean result of a MT-generated 32float-file when dropped into REW's RTA window, don't you? It is basically impossible to get a different result (other than a somehow broken version of REW).
This proves the file itself is OK (as is another one created by REW, this time). And there aren't any soundcards involved yet.

Now, using that file in MT for actual loopback testing, it tests artifact-free, provided that the output driver is WASAPI/ex and the play gain is not exactly 0dB.

- the file itself is clean
- load and direct play as is in MT yet shows errors even with WASAPI.
- changing the gain "fixes" it.

The 32-bit fp file generated by REW seems OK, at least when analyzed by DeltaWave it shows no artifacts. This is the same code that reads the file in MT, so file itself and reading it into 64-bit floats internal to MT seems OK:

 

[KSTR]

The 32-bit fp file generated by REW seems OK, at least when analyzed by DeltaWave it shows no artifacts. This is the same code that reads the file in MT, so file itself and reading it into 64-bit floats internal to MT seems OK:

View attachment 223061

Fine.
So, atm we have
- exported 32bit file is OK. Both MT's and REW's.
- import to MT and convert to 64float is supposed to be OK as the same code is used in DW.
- local 50Hz preset generates data directly in 64bit domain (I suppose) and always tests OK when using WASAPI/ex.
- but loading the above 50Hz preset that was saved in 32bit into MT fails with 0dB playback gain even with WASAPI. Changing gain fixes it (but not with ASIO output).

What is exactly happening when applying play gain in MT, do you (re-)dither the loaded file content? That could be the key point as I see it because the gain process appears to fix the artifacts for whatever obscure reason.

 

[Grooved]

@Grooved , are you using ASIO?
Because I see what I call the "conversion artifact bug" in all of your four plots, the one I've shown above in my plots when using ASIO.

24bit with no dither should have looked like this:
View attachment 223011
The -300dB "holes" at certain FFT bins are a good indicator whether undithered data gets handled properly.

And 24bit with dither it should look like this:
View attachment 223012
The dithering evens out the baseline noise floor as expected. Harmonics are again all below intrinsic processing noise.

I can only get this perfect pureness with WASAPI/exlusive.

Hi @KSTR ,

thanks, and yes I was using ASIO but with FLEXASIO and ASIO4ALL as it was impossible to select the digital channels in Multitone.
These ones were made with FLEXASIO and WDM-KS setting, as I think I got better results than with WASAPI (exclusive or shared)

I will try again with REW where we can select any channels with the ASIO driver

 

[KSTR]

And as a general remark, I'm just curiuos what's going one, @pkane. Should you decide it's a "don't fix" microscopic issue I'm completely fine as I know the workarounds by now.
Anyway, a big Thank You again for this brilliant software. The arbitrary FFT size is a killer feature for me.

 

[bennetng]

I'd suggest using 6kHz sine with 48kHz sample rate, any error if visible should be at 18kHz only.

 

[pkane]

And as a general remark, I'm just curiuos what's going one, @pkane. Should you decide it's a "don't fix" microscopic issue I'm completely fine as I know the workarounds by now.
Anyway, a big Thank You again for this brilliant software. The arbitrary FFT size is a killer feature for me.

I'm curious about this, as well, just don't want to spend too much time on it now, while there are much more serious issues and bugs to be fixed