Jump to content
IGNORED

DeltaWave null-testing audio comparator (beta)


Recommended Posts

2 hours ago, TomCapraro said:

The electrical conditions appear to be under the same clock, with the same cables, and with the same ADC that recorded them.

 

Yes, I was thinking it would be interesting to experiment with changes to cables, various electrical noise reduction strategies (using or not using particular power supplies, for example), and see whether anything at all came of it.

One never knows, do one? - Fats Waller

The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein

Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature.

Link to comment
11 minutes ago, Jud said:

 

Yes, I was thinking it would be interesting to experiment with changes to cables, various electrical noise reduction strategies (using or not using particular power supplies, for example), and see whether anything at all came of it.

Of course something comes out
In the archive I have acquisitions made on an original OPPO CD player and then after changing the clock, the capacitors and the power supplies are separated.
I look for them tomorrow and post the results.

Link to comment
3 hours ago, pkane2001 said:

 

Frank, trying to reproduce the error with narrow-band drift correction. Are you testing with original 44.1k files, or upsampled ones?

 

 

Upsampled. Note, alternative drift method invoked, which is likely to be the main reason,

 

Marley47.thumb.PNG.e497c30377e5f56d5c773c41eb2003a3.PNG

 

Marley48.thumb.PNG.36c13e4d8d329499f2b834934e1bb735.PNG

 

Good offset, poor drift compensation.

Link to comment

Captured a Manual window misbehaviour - the sequence of operations is key, but I can't say yet what the exact combo is,

 

Marley49.thumb.PNG.5d6be1f1cb9ddaca937f3370107acad5.PNG

 

In the first adjustment Cache Data was not checked, but was for the 2nd - previously, it was the other way around ... I think!

Link to comment
On 4/24/2019 at 10:53 PM, Jud said:

 

Yes, I was thinking it would be interesting to experiment with changes to cables, various electrical noise reduction strategies (using or not using particular power supplies, for example), and see whether anything at all came of it.

451/5000
 
 
 
Given the result of the null test relating to the OPPO 205 device to which the capacitors were replaced, the clock improved the power supply.
Nothing could be seen of traditional measurements, with Deltawave one can see a difference.
The graph shows the two Delta of spectra that derive between the original signal (pink noise) and the analogue stage signal.
In blue the OPPO 205 improved, in white the same device without modifications.

ktm.jpg

Link to comment
18 minutes ago, TomCapraro said:
451/5000
 
 
 
Given the result of the null test relating to the OPPO 205 device to which the capacitors were replaced, the clock improved the power supply.
Nothing could be seen of traditional measurements, with Deltawave one can see a difference.
The graph shows the two Delta of spectra that derive between the original signal (pink noise) and the analogue stage signal.
In blue the OPPO 205 improved, in white the same device without modifications.

ktm.jpg

Hi Tom,

Was this difference perceived in listening tests too?

Link to comment
5 hours ago, bibo01 said:

Hi Tom,

Was this difference perceived in listening tests too?

hi Gianni, at the time a listening test was conducted (with dufay and others, it seems there was also the participation of Emidio Frattaroli) and it seemed to me quite reliable. Most blind participants chose the modified OPPO reader.

Link to comment

Doing another round with the Bob Marely orig and 1st gen; phase shifts from the electronics path, DAC through to ADC, are what's keeping better nulling happening - trying to work a way around this, and ended up with using a 12000 HP@ start filter to examine something. Which showed up some issues ...

 

In 44.1k DW was completely incapable of synching, and reported this. But, using 176.4k resamplings of same, a very good result was achieved; matching was fine, and a worthwhile null achieved ... could DW do this internally, to match the manual workaround?

 

DW is not handling Cache File Data selected well, possibly because that filter was selected - I didn't check; the way out is to deselect, and the results then make sense.

 

When using the Manual window, it would be very useful if the chart that's being displayed, at a zoomed in detail level, maintains that zoom after Apply clicked, and the new results displayed. Otherwise, axes are reset, and one has to go through a ritual of zooming to find the same spot - to see what impact there was in possibly a key area.

Link to comment
19 hours ago, fas42 said:

Doing another round with the Bob Marely orig and 1st gen; phase shifts from the electronics path, DAC through to ADC, are what's keeping better nulling happening - trying to work a way around this, and ended up with using a 12000 HP@ start filter to examine something. Which showed up some issues ...

 

In 44.1k DW was completely incapable of synching, and reported this. But, using 176.4k resamplings of same, a very good result was achieved; matching was fine, and a worthwhile null achieved ... could DW do this internally, to match the manual workaround?

 

DW is not handling Cache File Data selected well, possibly because that filter was selected - I didn't check; the way out is to deselect, and the results then make sense.

 

When using the Manual window, it would be very useful if the chart that's being displayed, at a zoomed in detail level, maintains that zoom after Apply clicked, and the new results displayed. Otherwise, axes are reset, and one has to go through a ritual of zooming to find the same spot - to see what impact there was in possibly a key area.

 

Thanks, Frank. I thought about adding auto upsampling as an option early on. Sometimes it's just a question of the number of available samples, as DW makes some assumptions about what size chunks of data to use for cross-correlation analysis. If there's not enough, it may simply not work well when computing drift. Upsampling creates more data for DW to look at, even though it's really the same data, just interpolated. The right way to handle this is to make the chunk size variable, adjustable to the number of available samples. Or, maybe upsample to simply generate more data :) I'll play around with these options.

 

Was it a specific chart that kept resetting scale? Generally, the charts should keep their scale across different manual adjustment attempts.

 

 

Link to comment

Sorry Paul, I haven't double checked what was happening, yet ... :/. Will aim to try to reproduce, soon ...

 

In the meantime, something I found interesting comparing the Marley orig, and 8th gen, over a bandwidth where the phase delay was largely constant,

 

Marley52.thumb.PNG.48c3827cc039db0972c984b63b026c71.PNG

 

This implies that the converter chain, over many generations of use, is almost sample by sample altering the effective gain, making any sort of decent nulling almost impossible - is this a significant aspect of audible distortion, or nothing of concern ... hmmm ...

Link to comment

And just found a factor that seems to be relevant, as regards losing the scale setting - if the zoom was not done immediately prior to applying an adjustment, then DW can lose the knowledge of it; doing a 'dummy' zoom beforehand forces DW to respect the setting.

Link to comment
11 hours ago, fas42 said:

And just found a factor that seems to be relevant, as regards losing the scale setting - if the zoom was not done immediately prior to applying an adjustment, then DW can lose the knowledge of it; doing a 'dummy' zoom beforehand forces DW to respect the setting.

 

Right. That’s by design: until you change the zoom level manually, the charts will continue to auto-scale to the new data. Once you pick a desired zoom level, they should stay there.

Link to comment

It must depend on the particular actions done, in between adjustments, or at the start - when I had the annoying behaviour, I would do an adjustment, lose the scaling; so, pan and zoom in again, do another adjustment, lose the scaling, etc.

 

I'll wait for the next occasion, and try to get a better handle on it ...

Link to comment

Happened again, just then ... was it to do with locking, unlocking? No, not obviously so - seems to be around the windows losing focus; if you do something else not directly related to DW, then the problem arises, possibly ...

Link to comment
2 hours ago, fas42 said:

Happened again, just then ... was it to do with locking, unlocking? No, not obviously so - seems to be around the windows losing focus; if you do something else not directly related to DW, then the problem arises, possibly ...

 

Seems a bit unpredictable, Frank. Let me try to reproduce it here.

Link to comment
52 minutes ago, pkane2001 said:

 

Not familiar with Solo concept in Audacity, but I like the idea :)

 

That is a good idea from Frank.  Yeah, you can have multiple tracks open in Audacity.  If you hit play, they all play.  If you solo one, then the others are muted. So being able to engage a solo function where you click and immediately switch between Ref and Compare on the fly would sometimes be handy. 

And always keep in mind: Cognitive biases, like seeing optical illusions are a sign of a normally functioning brain. We all have them, it’s nothing to be ashamed about, but it is something that affects our objective evaluation of reality. 

Link to comment

Thanks, Dennis, for adding the extra detail.

 

Since the focus is now on Archimago's CD output test, I thought I'd apply DW to the tracks I listened to back then - and found not so good things ...

 

The track was the Joe Satriani - "Crowd Chant" one, and I compared the A and B versions, Motherboard vs. iPhone - what was obvious was that the captures were extremely different, so no wonder people heard things.

Spectrums matched, but the waveforms were radically distinct from each other - bad enough that DW failed badly at matching, and was 2 samples out in the offset - the delta had peaks half that of the reference! A single manual adjustment trial immediately corrected most of this.

 

Lots of phase shift upsetting things? I tried a bandwidth 0f 1000 to 8000 @start, and now DW was really struggling! Failed to detect drift, and prompted for alternative method - OK, but the matching was now a total disaster, and the delta level built to a crescendo by the end of the track.

 

So, it appears that this may be a good set of specimens of music to fine tune DW against, to improve robustness of processing.

Link to comment
13 hours ago, fas42 said:

Thanks, Dennis, for adding the extra detail.

 

Since the focus is now on Archimago's CD output test, I thought I'd apply DW to the tracks I listened to back then - and found not so good things ...

 

The track was the Joe Satriani - "Crowd Chant" one, and I compared the A and B versions, Motherboard vs. iPhone - what was obvious was that the captures were extremely different, so no wonder people heard things.

Spectrums matched, but the waveforms were radically distinct from each other - bad enough that DW failed badly at matching, and was 2 samples out in the offset - the delta had peaks half that of the reference! A single manual adjustment trial immediately corrected most of this.

 

Lots of phase shift upsetting things? I tried a bandwidth 0f 1000 to 8000 @start, and now DW was really struggling! Failed to detect drift, and prompted for alternative method - OK, but the matching was now a total disaster, and the delta level built to a crescendo by the end of the track.

 

So, it appears that this may be a good set of specimens of music to fine tune DW against, to improve robustness of processing.

 

Thanks for testing, Frank! Let's walk through your tests, to make sure I got everything right.

 

1. You compared Crowd Chant track A and B without doing any filtering on them. This produced a delta waveform with peaks of 1/2 of the original track, correct? What did you adjust that made it better? Offset, gain, drift?

 

2. When you band-pass filter 1k-8k range, you are obviously removing some important data. But, it may just be that the warning about using alternate method was a bit too aggressive.  I also got the same warning, but when I said 'No' to the prompt to use the alternative method, here's what I got:

image.thumb.png.eeb204b893d5ed787cb31ee0cc873cce.png

 

Seems like a pretty good result. I'll review what threshold to use for when this warning message pops up, as the logic was based on the poor precision of the offset calculation in previous versions. Can you please try answering 'No' to the alternative drift correction method to see if you get an improved result?

 

Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now



×
×
  • Create New...