I don't understand

[nemick]

I don't understand any of these graphs etc. that I'm seeing on this part of the forum.

 

Can some kind soul point me to a layman's explanation of what I'm looking at so that I can try to understand this stuff and hopefully contribute at some point?

 

[wgscott]

Here is a Fourier Transform (aka frequency spectrum) of a track from a DVD-A rip from American Beauty (one of the few Grateful Dead studio albums that doesn't suck). It was sampled to 96kHz, which means that you potentially can see a frequency response up to 48 kHz:

 

 

Notice how it tapers off very naturally as you approach 48kHz. That is a good thing, because it means there shouldn't be any artifacts due to imposing an arbitrary cutoff.

 

Here is one (posted in the Liszt organ thread) that is sampled at 88.2 kHz, so it should show frequency extensions naturally tapering off as you approach 44.1 kHz (twice the Redbook CD frequency range):

 

 

Notice in reality it is very abruptly chopped off at 22 kHz, corresponding to what a CD would have. This permits us to suggest that indeed it is simply upsampled CD (i.e., it is a fake high res file).

 

I think of this complementary view as a cross-section of the above, where false color intensity is now replaced with area under the curve. It shows very clearly there really is nothing beyond 22 kHz (maybe even 20 kHz):

 

 

Finally, the so-called waveform shows if the dynamic range has been exceeded. In this case (same Liszt track) it looks good:

 

 

If there is a lot of clipping, it suggests that the "loudness" has been artificially enhanced at the expense of maintaining a good dynamic range. Goldsdad found a nice example:

 

 

You can see that this was a deliberate decision if you compare it to the original CD:

 

 

 

 

[nemick]

It would be impossible for me to overstate how much I appreciate you taking the time and effort to show me this. It all makes a lot more sense now and I can see what the graphs etc. are telling me.

 

I've downloaded Audacity, so I'll have a go at trying this out myself.

 

Thanks again.

 

[Julf]

Here is an other example of a graph - this is how a SACD (DSD) source looks when resampled to PCM. Notice how the spectrum drops down beyond 20 kHz, and then the digital noise takes over. Anything above the dip is DSD noise.

 

 

 

[goldsdad]

http://www.oocities.org/frwrdevr/