[spruce music]

I am posting much the same thing 71 db has. Here is what I did. Created a pair of pulses at 192 khz rates that were 2 samples apart which is just about 10 microseconds. Then a pair 4 samples apart which is about 20 microseconds. I then resampled it to 44.1 khz. Then upsampled it back to 192 khz making it easier to compare. It looked like this. The 44.1 khz versions are slightly different. Please remember you are seeing sample values and not the actual reconstructed waveform. With high sample density they aren't much different, but with low sample density you can fool yourself this way.

Now back to the spectrum like 71 db is talking about. Let us look at the FFT of the 10 usec 192 file. It will be followed by the same thing after being sent down to 44.1 khz. Notice how the level below 20 khz is the same in both. Near 20 khz both are showing - 78 dbFS. The full range FFT shows the pulses are showing interference cancellation at 50 khz roughly. The low sample rate version only has spectrum to 20 khz, but it is the same as the below 20 khz spectrum in the high sample rate version.

Now the same FFT for the 20 microsecond pulse separation. Again both show the same spectrum below 20 khz, but in this case the cancellation interference is about 25 khz. And the 20 khz level is lower by 9 db and the spectrum is altered over at least an octave or more. So yes, these would sound different if you heard them as they have different spectra.

 

[spruce music]

Here I put a 13 khz sine wave in both channels, but offset by 2 samples at 192 khz. That would be about halfway between samples in a 44.1 khz sampling of them. I then downsampled this to 44.1 khz and upsampled to 192 khz again. Notice how the two channels are still offset exactly 2 samples in the 192 khz version. The timing difference even though it is only 10 microseconds between channels was preserved even at 44.1 khz. I also included the 44.1 khz samples at the bottom. Notice they are different between channels and remember you are seeing samples and not the actual waveform that gets reconstructed.

While all of these are software generated I have in the past done this for real with ADC and DAC using real signals. It works the same way.

I also wish to add two more things. There are problems with Kunchur's work. Others have shown 10 usec differences are audible. Kunchur claims half that, and uses that to claim higher rates are needed. What I have shown here indicates they are not needed for reasons of temporal resolution. Secondly single sample impulses are not legal signals in the sense of Shannon-Nyquist sampling. S-N refers to a signal passing thru an ADC and then DAC conversion. Single sample impulses would have gone thru an ADC with a low pass filter and would never end up sampled that way. So it is an artificial signal. Useful for studying filter response in some ways, but not something a music recording would ever have. Nor can limited bandwidth speakers or human ears hear an impulse as such a sharp singular peak, you only hear the below 20 khz part of it. As such peaks have infinite bandwidth or bandwidth as wide as the system can have which is trying to reproduce it.

 

[danadam]

@spruce music Did you do anything special to make your FFT window go that low on y-axis scale? I can't make mine go below -90 dB.

71 dB, spruce music, great explanations and examples. I can't wait to see in what ways the hires/mqa enthusiasts will dismiss or misinterpret them

 

[RRod]

@spruce music Did you do anything special to make your FFT window go that low on y-axis scale? I can't make mine go below -90 dB.

The Y-axis will change with the window type.

 

[spruce music]

@spruce music Did you do anything special to make your FFT window go that low on y-axis scale? I can't make mine go below -90 dB.

71 dB, spruce music, great explanations and examples. I can't wait to see in what ways the hires/mqa enthusiasts will dismiss or misinterpret them

In Audacity, under preferences, look in the Interface section. You can choose the depth of metering up to -145 dbFS.

Audacity however adjusts the lowest level based upon the lowest part of the signal shown. So even with a -145 db setting it might show only down to some higher level if there is no part of the FFT lower than that.

If you haven't updated Audacity in awhile, the newer versions do have a slider to the right on the FFT window that lets you zoom in and look at only the upper or lower portion of the FFT in more detail.

BTW, I used the Hanning window which is a good compromise most of the time.

 

[danadam]

In Audacity, under preferences, look in the Interface section. You can choose the depth of metering up to -145 dbFS.

That's it, thanks.

 

[Jimster480]

The advantage of Soekris over Schiit is native support of DSD.

DsD isn't very valuable though.... Personally 95% of audiophiles won't even use it.

 

[cheungtsw]

DsD isn't very valuable though.... Personally 95% of audiophiles won't even use it.

Not sure where your 95% come from.
Here in HK many albums will come with a SACD version and SACD version does sound better.

 

[Jimster480]

Not sure where your 95% come from.
Here in HK many albums will come with a SACD version and SACD version does sound better.

It's not scientifically possible.
DSD provides no advantages.
Also most people in Europe and the USA never purchase albums anymore since most people use streaming services.

I just read an article yesterday about higher bit rates actually introducing ultrasonics which can cause damage to your hearing via intermodulation while not providing any detail that you could ever hear.

 

[drwlf]

Agreed with @Jimster480
http://dr.loudness-war.info/ matters more, some SACDs suck, some don't, it's not universal.

 

[cheungtsw]

It's not scientifically possible.
DSD provides no advantages.
Also most people in Europe and the USA never purchase albums anymore since most people use streaming services.

I just read an article yesterday about higher bit rates actually introducing ultrasonics which can cause damage to your hearing via intermodulation while not providing any detail that you could ever hear.

You either are kidding or you dont know the hell you are talking about.

Anyway, no point to argue with you. Bye bye.

 

[m-i-c-k-e-y]

Why not judge it for yourself. Here is an old site to compare various format/depth/resolutions:

http://www.2l.no/hires/index.html

Problem only with 2L Recording is that their highest resolution for recording/mastering is DXD. We know our listening experience is only good at on how a material is recorded and mastered. Sound Liason does excellent recordings (check out their 24/96 materials). Euphora/Just Listen/Cobra Records etc. (nativedsd.com) now does DSD256 recordings and their materials are out of this world. Same too with Blue Coast which does analog recordings (2 inch tapes) and masters in DSD256. A bit pricey but very rewarding.

 

[Jimster480]

Well I don't just read, I listen. And I do hear the difference.

Why not judge for yourself. Here is an old site to compare various format/depth/resolutions:

http://www.2l.no/hires/index.html

Problem only with 2L Recording is that their highest resolution for recording/mastering is DXD. Euphora/Just Listen/Crier Records (nativedsd.com) now does DSD256.

No DSD will increase audio quality, already 16/44.1 is capable if our full audible sound recording with the exact accuracy in the time domain.
The rest is placebo affect.

 

[m-i-c-k-e-y]

Ha ha ha good luck on your 16/44.1.

 

[bigshot]

Welcome to Sound Science folks! Your ship has arrived on our shores courtesy of the powers that be.