| Author
|
Mythbusters: Audio Production Special
|
UnderTow
Started Topics :
9
Posts :
1448
Posted : Sep 5, 2008 03:15
|
Quote:
|
On 2008-09-04 18:42, Kaz wrote:
And we are talking about worst-case scenarios, hence, my comment is valid 
|
|
Oh you moved the goal posts. Interesting. 
Quote:
|
Incorrect, unless all frequencies are in the same phase
|
|
If a simple sine wave is very close to Nyquist, there will be (relatively) long periods of time where all the samples are either on the positive or on the negative part of the cycle. There is nothing in sampling theory that sates that the sample points have to be on the positive and negative parts of the cycle.
Quote:
|
- using your circle analogy, two points are enough only if you know for certain one is the center, otherwise, two points can give an infinite number of frequency/amplitude combinations.
|
|
You started with the circle analogy and jumped into sampling mid-sentence. I'll just address both:
Two points on the perimeter give two possible circles but we are talking about band limited signals (which is the key to the whole sampling theorem!) so there really is only one possible circle within the bandwidth.
Anyway, we always know the centre: In sampling, the centre of the circle (which is just a sine wave plotted over time) is the zero crossing. Tadaa!
(I've thought about this analogy or I wouldn't post it ).
Quote:
|
My bad, by doubling the sampling rate, hence making the interpolation twice as fine, hence, more accurate reproduction.
|
|
Again you mention interpolation. What interpolation are you talking about? In the DAC? If the interpolation is correct (for oversampled DACs), the sampling theorem holds. Adding sampling points within the limited bandwidth does not add accuracy.
Quote:
|
True, but again, we're talking about 100% system compatability, and most filters are far from standard equipment on say.
|
|
Compatibility with what exactly?
Quote:
|
That makes no sense, we are talking digital data - engineers don't care much for numbers like 1000, if anything - in such a case they'd work with 49152Hz (2048 samples per frame). It's nice to think that these standards make sense from a scientific point of view - unfortunately, they don't... not really.
|
|
They most certainly do (for most standards).
In 1981 48 Khz became the standard. It was agreed that the signal bandwidth needed to be at least 20Khz to take care of the human audible range. The current technology at the time for 16 bit quantization could run up to 60Khz. It was decided that the sampling rate should be a multiple of 600Hz (Which itself is a common multiple of the 24, 25, and 30 Hz frame rates commonly used in film and TV).
60 Khz would have been ideal but it was considered a waste by people that understood sampling theory! So they were left with leap frame frequencies (each video frame does not have the exact same number of samples) of the common video/film/carrier formats of 48, 50, 52.5 and 54 Khz.
There was already quite a bit of hardware and software on the market using either 48 or 50 Khz and In Europe, 32 Khz was used for broadcasting (again, for all sorts of technical reasons). 48 Khz had the advantage that it was a 3:2 multiple of that 32 Khz. It didn't quite fit in the NTSC format but that wasn't in use in Europe.
Quote:
|
Sure it is, set a 2xdowngrade in sampling rate and you will lose a lot of original character.
|
|
What is this character you speak of? It doesn't mean anything in a technical discussion. Maths please.
Quote:
|
Of course - but DSP architecture allows for cross-influence, hence working in double precision has no performance hit, as opposed to classic 32bit processors like GPUs. Programmability is a wonderful thing, and that's the first feature that the high end plugins for these cards use (at least in any plugin that uses the word "convolution" or "emulation" somewhere).
|
|
Goobledegook. English please. (Throwing in pseudo-technical terms without any real explanation won't do ).
The word BUT usually means you are going to write something contradicting my point, not supporting it. Well it was your point originally but I am agreeing with it.
UnderTow
|
|
|
UnderTow
Started Topics :
9
Posts :
1448
Posted : Sep 5, 2008 03:26
|
On 2008-09-04 19:14, Kaz wrote:
Quote:
|
What you're saying is not right. It's not even wrong.
|
|
Heh?
Quote:
|
I'm just saying there's a different set of problems in digital and in analog processing, and gave this theoretical bit of analog design philosophy -
|
|
So why not stick to the theoretical aspects of digital audio? Anyway, amplification will not in any way solve the problem of the noise floor. Neither theoretically or practically.
Quote:
|
as this is an out of context reply as to why sampling rates and bits simply do not exist in the real world... the sound we hear is a vibration of air, not numbers.
|
|
Those vibrations can be measured.
Anyway, the answer was perfectly in context to your comment of "the dynamic range of an analog signal is limited only by amplification" which is still rubbish.
Quote:
|
No, it's not. While DACs work in MHz, it's still just a lot of dots before it becomes an electrical signal - as long as we're working with samples, it's digital information. Once it is an electrical signal, it is analog (by definition, it is not digital information anymore).
|
|
Errr... What are you arguing about? I wrote: By the time a digital signal is converted back to analogue, it is a continuous signal again. Notice the word analogue in that sentence.
Quote:
|
DACs use extremely high-speed calculations into making sure the electrical signal is a very accurate representation of the digital signal - and it's closer to linear work than in the PC, but it's not linear-based, it's still sample based on the digital form.
|
|
heh? Besides using the term linear incorrectly, what does that have to do with my comment?
Anyway, the reason ADCs and DACs use oversampling has to do with engineering issues. Not with sampling theory. The signal is accurately represented at the base rate or oversampling the DAC wouldn't have any benefit would it? You can not generate more accuracy than you already have.
UnderTow |
|
|
|
UnderTow
Started Topics :
9
Posts :
1448
Posted : Sep 5, 2008 03:39
|
Anyway, now to the main reason for me joining this topic:
Although your intentions are commendable and your knowledge is way above average, you can not start a topic about myth busting if you do not address the issues with absolute clarity, complete lack of ambiguity and an entirely correct use of terminology. Elsewise you will just end up adding more confusion and creating a few new myths to boot!
UnderTow |
|
|
|
|
