the quote is a LOT more helpful and specific! The variables are noted.
I didnt use a lab that suggested what to do with the data nor doctor that suggested any such application. Thus, no quackery. This report however, is not unlike many other lab tests - lab to lab - even for blood tests are not identical. Even engine Dynomometers vary on the output results within the same maker and esp between brands which may use different methods to derive the answeres. that does NOT make any of them useless nor quakes. Of course, more than just this one test is needed to get a full and more complete picture.... even saliva tests.
Thanks for the more direct and less confrontational info. -RNM
Last edited:
But you don't have to. The looping thing is a conceptual aid for those who don't get the idea that ANY continuous time domain signal of less than infinite duration can be Fourier transformed.
So they claim. From the literature I've seen, there's a lot of question marks. I wouldn't ever make my employees do that.
You still don't get it, do you? A CR circuit (assuming it is made of linear components, which may include DA) does not add any frequency components! It merely adjusts the amplitude and phase of the components in the input signal. It does not add new frequency components. Sampling theory is completely irrelevant, as no sampling is taking place.simon7000 said:
A single piece of music might not be periodic, but any section of it is likely to be approximately periodic. Repeating it makes it periodic. Unless you are claiming that repeating a track will make it sound different through a CR filter than playing it once then we have established perodicity, and we know that music has no discontinuities so Fourier works. No new frequency components, so no non-linear distortion. All we have is a tone control! It really is that simple. I am baffled why you find this so hard to accept. Why continue to assert something which is simply not true?
Yes, Fourier transforms work. But if pauses in the loop are of different length results will be different.
I band limit on the low end by turning my recorder on then later off. If you want we can band limit on the high end well beyond 20k and sample at a higher frequency. In either case this process does not make the signal "periodic". I was talking about a single recording not looped. At 44.1k we can get about 95 seconds in a 4Meg file, easy to play with in Octave or the free MathCad that comes with the book.
DF96 is right, I don't know where this processing real music through an R/C makes signals at new frequencies came from.
FBI forensics, for one, think so. I mean they claim to think so. maybe they are just guessing or are being fooled by quakes.
Now look what has happened -- I'm doing it :-( I have caught the 'cute' virus.
Last edited:
Not relevant to the claim of new frequencies being created. Windowing is your friend in general, but in this case I was simply stating that a set of time domain data and its FFT are a unique reversable process. Using double precision math 24 bit original data should be exactly recovered with a simple ceiling or floor roundoff (forgot which one).
I have not replied your post. Anyway, any time domain signal of finite length has infinite frequency spectrum.
I do remember reading Omni magazine around 1980. If it was Vernor Vinge’s work, I can’t tell.
I would like very much to confine myself to technological issues.
But history is a source of worry as to what sick minds do with every powerful outcome of science and technology. (Excuses were always available).
Technological singularity - Wikipedia, the free encyclopedia
Me, I naively point to Asimov’s three rules.
George
SY,
Lets take a flute sounding a note. This was one of the sounds that gave early electronic music systems problems. It starts out as air noise and then begins to resonate around a note and sort of settles in and then finally fades out. If you want to use Fourier analysis on that you do need to loop it. (To get useful practical data.)
Now if you do an analog recording of it, you add noise, some distortion and bandwidth limiting to mention the most common problems.
When you record it digitally there are sampling issues that arise. These are well correlated with bandwidth limits.
Scott,
As an example we may have a signal such as noise that we may model as a random series of unit impulses. Each unit impulse can be considered to have a very large number of harmonics. (limited in practice not theory.) When they pass through an RC filter the harmonics will have changed in level and phase. Now a change in harmonic levels most consider harmonic distortion.
As there are in theory infinite harmonics to that extent there can be no new ones.
Now is there a non-periodic signal that does not have infinite harmonics?
Last edited:
No, not really. That may help s/n, but in these days of modern times, it's trivially easy to get more than enough s/n in a single shot. Back in the early synthesizer period, getting 24 bits, nanovolt noise floors, and multihundred kHz sample rates was unimaginable.
I see the root of this, the above is conceptually wrong. Creation of new frequencies by a non-linear transfer function is harmonic distortion. Purely additive or substactive processes are not non-linear.
Did Pavel mention a symphony? I think he said "any time domain signal of finite length has infinite frequency spectrum" and he is right.
Cheers
Stein
- Status
- Not open for further replies.