For such an understanding/interpretation, any undergraduate EE student would catastrophically fail his Signals and Systems exam. It's of course his professor's fault of not knowing about trigonometric identities
.Sorry JN, don't you think you made enough of a fool of yourself? This damage may remain permanently on your forum record.
The result with info on the cymbal listening test is now posted, head over to:
Poll: Cymbals of different sampling rates listening test
Poll: Cymbals of different sampling rates listening test
Impossible is the key word, the concept of filtering does not directly apply to the product of two signals, the convolution must be performed first to get a simple sum of frequency components, the frequencies in the trig identity are mathematical place holders.
It will be noted that 65% of the participants do not perceive any difference, and that 0% of the participants gave the correct answer, even by reversing their imaginations of what could be the original and the duplicated copy.
Sigh...
Here is a comparison of the waveforms between the sum of 17.5 Khz and 22.5 Khz (in blue) and the product of a 2.5 Khz sine and a 20 Khz sine (orange).
I had to truncate the product (orange line) from zero to 200 points, and 800 to 1000 points so that it was evident that a blue line was even on the chart. I guess I could have shifted one up or down a few pixels, but then someone would claim they weren't identical.😀
Needless to say, these two techniques for producing that waveform prove that the trig identity holds.
It also proves that multiplying a 2.5Khz sine times a 20 Khz sine is indistinguishable from the summation.
It also shows that an amplitude modulation using two in band frequencies (2.5 k and 20) will generate a signal at 22.5K.
jn
Here is a comparison of the waveforms between the sum of 17.5 Khz and 22.5 Khz (in blue) and the product of a 2.5 Khz sine and a 20 Khz sine (orange).
I had to truncate the product (orange line) from zero to 200 points, and 800 to 1000 points so that it was evident that a blue line was even on the chart. I guess I could have shifted one up or down a few pixels, but then someone would claim they weren't identical.😀
Needless to say, these two techniques for producing that waveform prove that the trig identity holds.
It also proves that multiplying a 2.5Khz sine times a 20 Khz sine is indistinguishable from the summation.
It also shows that an amplitude modulation using two in band frequencies (2.5 k and 20) will generate a signal at 22.5K.
jn
Attachments
Last edited:
Who is the other brand? Whose DSP is in it?
Minidsp has a lot of different products some with only digital in and out. They are all based on the ADI Sigma Studio stuff which is really flexible. It's also the model the other dps guys are emulating.
You continue to miss the point. However a waveform is created in the first place doesn't matter. Once it is created it is only one entity, in the case of your example there are exactly two frequencies resulting. That's the answer.
We agree on that, its not a sticking point.
Please give an example of what is being multiplied by what in the case of an analog to digital converter with preceding brickwall filter. Only frequencies below fs/2 survive through the filter. Do we agree on that?
Taking it from there, where and how do any processes of addition, multiplication, interference patterning occur? If you say multiplication occurs, how do you personally distinguish if something is multiplication or interference?
Last edited:
Without pausing for a nanosecond and questioning the highly probable glitch in graphic representation of waveforms by one particular computer application (LTSpice). Others have pointed this out but it apparently fell on a def ears.
No, I don't... you miss it.
How's that for talking past one another...😀
My exact point.
For this simple example, yes.
Absolutely. The example of modulation I have provided is a simplistic one so that everybody can understand (boy, was I incorrect in that regard)
Music. I can post an example in a few minutes, wait...
For the cited example, it was already stated that they are indistinguishable.
here is a ridiculously simple representation of a bell being hit. I modulated the fundamental frequency by 1/x. (and yah, I should have increased the fundamental frequency at least a factor 5 to put a lot more wiggles in..)
The leading edge of course totally violates nyquist. The trailing edge, the steep part could depending on the fundamental frequency, the tail, nah.
jn
Attachments
Last edited:
A glitch in graphical representation on the exact same plot, where the output waveform in two "proofs" were time expanded almost the same amount, and that glitch didn't affect the waveform it was overlapping?
Do you really want to stick with that?
Besides, that was the first thing I said, someone else needs to verify.
What about you?
jn
Last edited:
Yep, nobody could with full certainty tell apart cymbal recording at 24/176 vs 320kbs mp3, perhaps the reason some cymbal recordings sound "blurry" is rather due to the microphone being put in a dark spot, or what does T with experience of recording industry think? 😀
Yes! Any half way decent researcher would fire up an alternative application or check it in some other way to see if the results match.
Given that you have the resources, where is yours?
At least KSTR tried it, he provided the exact same results as Hans.
jn
I would assume that the location of the mike at any position in space, either over or under, could be used to change the sound.
jn
- Status
- Not open for further replies.