Simon Barnes wrote:
>happens exactly with mathematics, but of course, you can't listen to that.
A
>22.05 kHz sinewave WILL come out of the D/A as a square wave (if you
>disregard oversampling, and the D/A slew rate) and needs to be filtered to
>remove harmonics above 22.05 kHz if you want to reconstitute the sinewave.
You have to remember that the D-A convertor is working on the assumption
that the signal to be recreated does not contain any frequencies above half
the sampling rate. It is up to the recording equipment to ensure that this
condition is met, to avoid the nasty aliasing that will occur as a result of
attempting to digitize anything above this critical frequency. And what D-A
convertor doesn't use oversampling and gentle slope analog filtering anyway?
Next, consider what a square wave is. It's a sine wave with truckloads of
harmonics. For a 15KHz square wave, the first harmonic is at 30KHz and will
be filtered out, as will every other harmonic. What this means is that
effectively any waveform above 10KHz will be reproduced as a sine wave,
regardless of what shape it was originally. This is not limited to digital
recording, but indeed any recording medium will behave the same above half
its frequency limit.
What is the consequence of all this? Absolutely nothing because we can't
hear the harmonics that make a 10KHz square wave sound different to a sine
wave. Try it with a function generator running through your stereo. As
long as the sine and square waves it generates are at the same level, you'll
notice that above probably about 6KHz the difference between them will
diminish and eventually disappear.
BTW, oversampling in a D-A convertor gives back the redundant data that
wasn't recorded on the CD in the first place (and would be on a DVD-A) with
the same accuracy as the data on the DVD-A. The only condition to the
success of this interpolation process is that no sounds above half the
sampling frequency are present in the digitally represented sound. This
condition also applies to DVD-A, but the frequency limit is higher.
I guess the point of all this is that mathematically the entire human
hearing range can be faithfully recorded within the constraints of CD audio,
and using higher sampling rates doesn't improve the accuracy - that is a
limitation in the A-D conversion before we even buy the disc. There is no
reason to believe that higher sampling rates will be used during the
mastering of DVD-A, because oversampling during A-D conversion pushes the
actual sampling rate far higher than the final media contains anyway. The
limit of usefulness for oversampling is a function of analog filter slope
and the highest desired recordable frequency. This will equate to the same
actual sampling rate for CD and DVD-A, because the final receiver of the
signals - human ears - dictates a highest frequency of around 20KHz.
I hope this all makes sense.
-cb
-cb
-----------------------------------------------------------------
To stop getting this list send a message containing just the word
"unsubscribe" to [EMAIL PROTECTED]
