DeltaWave null-testing audio comparator (beta)

[pkane2001]

26 minutes ago, TomCapraro said:

120/5000

 

 

 

Don't worry ... how is the software for multitone measurements?
Have you completed the program?

Yes, the software was designed to measure devices using multi-tone signals. It's functional, but not ready for others to use, quite yet. At some point I may publish it, when I have a little more spare time. Here's an example result using a 20k multi-tone  Blue is the full test signal, white is the distortion with the test signal notched-out:

 

 

The program computes clock drift, frequency response, and TD+N for any number of multi-tone signals, from 3 to thousands. I was also planning on computing phase response, but didn't get to it yet.

[TomCapraro]

9 minutes ago, pkane2001 said:

Yes, the software was designed to measure devices using multi-tone signals. It's functional, but not ready for others to use, quite yet. At some point I may publish it, when I have a little more spare time. Here's an example result using a 20k multi-tone  Blue is the full test signal, white is the distortion with the test signal notched-out:

 

 

The program computes clock drift, frequency response, and TD+N for any number of multi-tone signals, from 3 to thousands. I was also planning on computing phase response, but didn't get to it yet.

😃😃😃

[TomCapraro]

I had read it on audio sciencereview.
I hope you will not be long in making it downloadable because if you waste too much time you will have to send me a provisional version to start trying

[TomCapraro]

91/5000

 

 

 

Paul, I'd like to try your 20,000-line multitone signal.
Can you at least send me this?

[TomCapraro]

In case you decide to send it to me please create two, one with 20,000 tones at 32bit / 44.1khz and another with 20,000 tones at 32bit / 96khz.

[pkane2001]

6 hours ago, TomCapraro said:

In case you decide to send it to me please create two, one with 20,000 tones at 32bit / 44.1khz and another with 20,000 tones at 32bit / 96khz.

 

Here are the files, Tom:

 

https://www.dropbox.com/sh/ieyurbgbndvjwkz/AADlqUEdxceasMJC3Q2MEYUNa?dl=0

 

60 seconds each, 44.1k and 96k @ 32bits.

 

44.1k one contains 4178 tones, and 96k file contains 5459 tones. That's because my software automatically removes any tones that are too close to each other (within a few Hz). This gives the notch algorithm enough space between tones to remove the original ones. Oh, and the start frequency is 14Hz.

 

Let me know if you find something interesting to do with these 

 

[TomCapraro]

8 hours ago, pkane2001 said:

 

Here are the files, Tom:

 

https://www.dropbox.com/sh/ieyurbgbndvjwkz/AADlqUEdxceasMJC3Q2MEYUNa?dl=0

 

60 seconds each, 44.1k and 96k @ 32bits.

 

44.1k one contains 4178 tones, and 96k file contains 5459 tones. That's because my software automatically removes any tones that are too close to each other (within a few Hz). This gives the notch algorithm enough space between tones to remove the original ones. Oh, and the start frequency is 14Hz.

 

Let me know if you find something interesting to do with these 

 

I will definitely keep you updated.
Thanks a lot for the files.

[TomCapraro]

If you send me two more files of 1 minute with the same frequencies (32 / 44.1khz and 32 / 96khz) but this time with 32 tones I will do some interesting tests.

[TomCapraro]

meanwhile I'll show you the first tests in loopback (PC + realtek ALC887)
The first with a maximum level of 0dBFS
The second with -0.2dBFS level

Frequency 44,1khz.

[pkane2001]

5 hours ago, TomCapraro said:

If you send me two more files of 1 minute with the same frequencies (32 / 44.1khz and 32 / 96khz) but this time with 32 tones I will do some interesting tests.

 

Alright. Added two more files. These are named mtone32_...

 

https://www.dropbox.com/sh/ieyurbgbndvjwkz/AADlqUEdxceasMJC3Q2MEYUNa?dl=0

 

Looking forward to more interesting tests

 

[TomCapraro]

Is it normal that the spectral lines are 27 rather than 32?

[pkane2001]

5 hours ago, TomCapraro said:

Is it normal that the spectral lines are 27 rather than 32?

 

That's because my software is trying to space out all the tones for easier removal after the capture. All the tones that are rejected are between 14Hz and 22Hz, so you're not missing much. If you want, I'll post the files with exactly 32 tones. 

[pkane2001]

5 hours ago, TomCapraro said:

Is it normal that the spectral lines are 27 rather than 32?

 

Try the new mtone32_*  files I just uploaded. Should have exactly 32 tones this time.

[TomCapraro]

That's fine Paul.
See you soon and thanks.

[TomCapraro]

122/5000

 

 

 

A curiosity Paul, the program with which you generate the tones ... can it also generate a multitone starting from a musical signal?

[TomCapraro]

Paul, in your opinion: why doesn't deltawave totally cancel the tones?
According to a certain logic, the cause should concern the alteration of the analog tone compared to the original tone? ... so that the residue of an altered analog tone would remain?

[TomCapraro]

148/5000

 

 

 

... I also tried to record the analog multitone signal twice in succession.
Again, the tones are not completely erased.

[TomCapraro]

...I think that I have understand.
The use of periodic multitones was enough to reduce the problem down to the base.

[pkane2001]

6 hours ago, TomCapraro said:

...I think that I have understand.
The use of periodic multitones was enough to reduce the problem down to the base.

 

What do you mean 'periodic multitones'? What did you do to make them subtract perfectly?

[TomCapraro]

2 hours ago, pkane2001 said:

 

What do you mean 'periodic multitones'? What did you do to make them subtract perfectly?

Hi Paul, to create a periodic multitone signal (whose visualization on FFT does not require weight windows, so to use with a rectangular or dirichlet) you have two solutions:

1) use the REW multitone signal generator, then set to 64k and at least 40 periods on 44.1khz.
Eventually the file will be 59.443 seconds long.

2) or take your multitone signal, load it into a good editing software, then cut a piece with 65536 samples and paste them so that 40 frames are formed.
Also in this case the files will be 59.443 seconds long.

This is your signal transformed into a periodic multitone.
Deltawave, as discussed in the past, works excellently with periodic signals, facilitating their alignment and accuracy.
Under normal conditions, the tones will have to cancel out ... but not down to background noise levels.
A small part will have to "emerge" but ... it will have to have an amplitude equal to about the same amplitude as the harmonic distortion with narrow band signals (such as a fundamental at 0dBFS)
In practice the residual of the multitones will correspond to approximately the level of the harmonics of the THD.
If, on the other hand, the residual is too high ... there is an error.

32 TONI periodici.wav

[TomCapraro]

Here is where the tone is canceled.

[TomCapraro]

Here, in my opinion, is the area that ... being non-linear (in addition to undergoing continuous variations as it is very close to the thermal agitation of the noise), it cannot be canceled.
It also seems normal to me as the base of each tone can also contain phase jitter.
At this point, the more the tones are canceled ... the more precise the D / A - A / D acquisition system.

[TomCapraro]

In overlay I added (white line) the original tone.
This thing, if you think I haven't shot a bunch of bullshit 😄, does your deltawave software credit because it perfectly centers where the non-linear part of the tone begins.

[pkane2001]

Folks,

 

After a delay, there's now a new version of DeltaWave available for download and testing: v1.0.51.

 

This version should improve speed, add a few enhancements, and produce better, more predictable and stable results, even with higher sampling rate files. The changes are as follows:

 

• Change to automatically pick a better size FFT for cross-correlation
• Change to reduce the number of iterations when computing drift
• Added option to change the file trim end setting from how many to trim to how many to include (you can now set the starting point in the file and how many seconds to extract from that point)
• Added larger size FIR filter setting, up to 1M taps
• Fixed a bug that could result in a few zero samples being added at the very end of comparison file
• Modified memory management to improve handling of files that are larger than a couple of minutes

Please give this version a try!

 

[fas42]

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, 🙂.