Are you saying sidebands cannot be created without a nonlinearlty? I've looked for, but can't find, evidence to the contrary.
Did you find any evidence of envelope modulation in musical instruments creating sidebands?
My thrust was more in terms the live-to-reproduced simulation. Namely, let's assume an extremely intimate venue/absolutely front row seats where you're sitting 10 m from the performers: propagation loss @ 20 C / 45% RH at 20 kHz is 5.5 dB; 30 kHz is 9.5, which I think is safe to call a substantial attenuation. Close mic'ing, unless patched through an appropriate set of filters which (drumroll) roll off the higher frequencies, like, say, an antialiasing filter.
The greater point I'm making is that for any of a hundred thousand reasons stereo playback is its own highly processed medium, not remotely an accurate simulation of anyone's prior experience. So there's a practical question of whether this entire dragged out back and forth has the slightest grounding in even one person's experience. That is not to say that one could prefer a certain type of synthesized music that pushes the limits of what can be recreated in 44.1/16.
None of this argues against the value of higher sampling rates in the transfer medium, as it gives more room between fs/2 and where music/humanly perceptible audio peters out, but it'd be really really nice if people could step away from their perfect mathematical transforms about fairies dancing on a pin to say whether this has a smidgen of practical value. Not even going to lengths of an audibility test -- close mic'ed with wideband microphones and high sampling rate, bang a cymbal or triangle, what's lost in the downsample to 44.1/16?
But you're getting modulation with a linear circuit. That's new, essentially self-contradictory by classic thought.
All the best,
Chris
The theory wants nothing above Fs. Since Fs is so low (in CDs), the filter is needed to attempt to eliminate the stuff above Fs since much music has stuff up there.
The filter can never be perfect and will introduce effects below Fs - phase perturbations, ripple, more depending on the filter used.
dave
What an odd ripple in the output. Looks to be about 3Khz.
I can't tell if it's something in your output, or a pixellation artifact. The other two do not exhibit anything like that.
The envelope is pretty slow as well.
Also, it's impossible to determine if the envelope modification is a consequence of the abrupt start, or it and modulation envelope slope.
That is why I specifically mentioned using a time mirrored exponential rise into a flat top, then exponential decay.
While you casually dismiss the alteration of the envelope by the filtering process, an audio researcher would probably not.
jn
I can't tell if it's something in your output, or a pixellation artifact. The other two do not exhibit anything like that.
The envelope is pretty slow as well.
Also, it's impossible to determine if the envelope modification is a consequence of the abrupt start, or it and modulation envelope slope.
That is why I specifically mentioned using a time mirrored exponential rise into a flat top, then exponential decay.
While you casually dismiss the alteration of the envelope by the filtering process, an audio researcher would probably not.
jn
This is simply not true but only a statement of limited available compute power.
What linear circuit does a bell use when I strike it and it exhibits exponential decay?
Did you just use a strawman?
jn
Here is some facinating info about the delay line inductors Quad 63 (and later) Delay Line Inductors let are around 2-3H air core inductors and quite lossy with a shorted coil inside.
I'll attempt a model for the delay line to see what it does. The model will be flawed for various reasons but still a start.
I tried some more tests with possibly more realistic signals. More importantly I actually listened to the 20k, 16k sum which had a 4k envelope and could hear nothing so I don't see how the presence or absence of the 24k could make a difference. I know at lower frequencies the signal envelope can be perceived.
It's the struck part he's alluding to; in steady state your ideal bell will have its fundamental and harmonics, but being struck we have a semi-step and decay modulation pattern multiplied (NOT added, where we'd see no new frequencies) on top of it, hence the additional harmonics.
Convolution in time = multiplication in frequency (vice versa)
I don't think I could hear that high either.
I also don't know how I would distinguish between a real cymbal in a room, one with CD brick filtering, and a hi res one. Apparently some believe they can.
jn
Some instruments are certainly nonlinear, I'd be fascinated to see the sidebands generated but the envelopes of different instruments. I can't find anything at all about it searching on line. The challenge, of course, with bells and cymbals, for example, would be separating the sidebands from the complex spectra.
I'm confident the closest I will get to the King of Sweden is a picture of him.
One day, I walked up to a women staring at a Nobel Prize picture, asked her if I could be of any assistance, answer any question...
She said, no, thank you. I was just looking at the prince, he grew up so much since I was there..
me:===>😱😱😱
jn
This confirms that the 4k beat envelope (clearly visible in time domain but does not show up in FFT) is inaudible.
Hi everybody! From what I gather in this discussion, CD (might) be compromised by difficult audio signals interacting with the anti-aliasing filter that is used, somewhere in the chain. How we fix this is to use 24-96K or even higher resolution recordings. Many quality recordings are made like this today, and that is what we seriously use for audio component evaluation or alternatively, a really classy vinyl playback system with at least a $2500 MC phono pickup. Everything else is too compromised to evaluate hi end audio systems.