24bit vs 16bit: How big is the difference?
Jun 13, 2008 at 1:43 AM Post #766 of 773
Quote:

Originally Posted by gregorio /img/forum/go_quote.gif
However, someone with say an undiagnosed heart condition could easily be killed.


So now you're a cardiologist, eh? LOL, sorry, but that's now how it works.
 
Jun 13, 2008 at 1:48 AM Post #767 of 773
Quote:

Originally Posted by gregorio /img/forum/go_quote.gif
This is because the noise is moved into areas where we can't hear it, leaving just the music.


Since you're dealing with 44.1 kHz audio, you can only move it in the 0-22.05 kHz band. You only have about 2 kHz lying outside the audio band, and there is no dither in existence that can stuff all the noise into that area. That means that if you use noise shaping, you will have increased noise within areas of the audio band. The average noise energy remains equivalent to that of the flat dither's -84 dB.

Quote:

The message is out of context


Don't misrepresent what I did. I didn't post a link to a message but to a thread, and there are a number of posts relevant in that thread.

Quote:

deals with self dithering during the 24bit ADC process and truncation effects during re-quantisation and could be misleading if applied to our argument of 16bit.


24 bit -> 16 bit is also requantization! The mathematics of dithering is the same; doesn't matter if you do it after the ADC or before DAC.
 
Jun 13, 2008 at 1:50 AM Post #768 of 773
Quote:

Originally Posted by Crowbar /img/forum/go_quote.gif
You're an idiot


pot... kettle... black.

See ya
Steve
 
Jun 13, 2008 at 1:53 AM Post #769 of 773
Quote:

Originally Posted by gregorio /img/forum/go_quote.gif
What goes into a DAC is a quantised digital datastream, what comes out has been converted into sine waves, that's why it's called a digital to analogue converter, duh!


The conversion from the stepped output of a DAC to a smoother signal is done by the analog filter. However, that just hides the lower resolution of a lower bit-depth word, sine if the steps are bigger, the filter in the time domain can be seen as doing interpolation. It still doesn't mean you have infinite detail--that is impossible since it implies infinite information density which is forbidden in physics. You're being purposely misleading.

Quote:

an analog signal that has been digitized can be perfectly reconstructed if the sampling rate was 1/(2W) seconds, where W is the highest frequency in the original signal." (Wikipedia: Nyquist-Shannon Sampling Theorum).


The mathematics of Nyquist assume non-quantized samples (infinite precision). This has already been pointed out.
 
Jun 13, 2008 at 1:55 AM Post #770 of 773
Quote:

Originally Posted by gregorio /img/forum/go_quote.gif
which have an infinte number of values because they are CONTINUOSLY varying!


They have an infinite number of values but resolution is limited because they are produced by the analog reconstruction filter from a finite number of output voltages the DAC chip produces! These output waveforms are continuous but there are finite number of them that can be produced.
 
Jun 13, 2008 at 2:01 AM Post #771 of 773
Quote:

Originally Posted by nick_charles /img/forum/go_quote.gif
Sorry, but this is wrong. No system with a finite bandwidth and a finite dynamic range can be other than finite. Continuous and infinite are not the same thing. A 16 bit DAC has a determined level of precision and infinite it cannot be, it may (or may not) do a perfect job of translating the supplied values into discrete voltage levels but it cannot for instance render a value represented as 1024.87654332 into a voltage as it cannot receive such a value.


It seems gregorio is confused because he's looking at the output of a DAC's analog reconstruction filter and assumes, incorrectly, that since the filter produces a continuous waveform that implies infinite resolution. That is, of course, false. It just converts the output signal from one basis to another. The basis functions of a DAC's output are step functions. The basis functions of the analog filter output are (approximately, since it's minimum phase not linear phase) sine waves. Low resolution in the DAC's staircase output is converted into low resolution expressed as increased smootheness of the output waveform. You're losing details either way.

You cannot have infinite resolution signal of any sort--even an analog signal is not infinite in resolution.
 
Jun 13, 2008 at 2:14 AM Post #772 of 773
[deleteme]
 
Jun 13, 2008 at 4:55 AM Post #773 of 773
I've locked the thread. 52 pages to read through to figure out exactly where I have to start deleting to clean up the mess. It will reopen at a later date.
 

Users who are viewing this thread

Back
Top