Phil Leigh;630361 Wrote: > The "distortion" does not rise with frequency. I think what you are > missing is: > a 20kHz signal never occurs in isolation in real music and even if it > did, > it takes many consecutive samples to convey a "sound" > there is no "averaging", there is a series of instantaneous samples (16 > or 24-bit words) replayed in sequence, in isolation > The averaging occurs only in our brain.
It seems to me the distortion has to rise with frequency. At least the part of it affected by dithering. At higher frequencies, there are fewer samples to average together to realize the benefits of dithering. Actually, the averaging occurs in the reconstruction filter, whether it be analog or digital, not my brain. This is mentioned on the page in the link. Look at Figure 8.1 in the link. It is based on a 4 bit sampling, so would be the same as an audio signal about 72 dB below full scale. It is also pretty low in frequency. I count well over 100 samples in the period of the damped signal, so it would have to be about 400 Hz. So try to imagine this same figure with 1/10th the number of samples. That would be the situation for a 4kHz signal. What would the plot in 8.1d look like then? Obviously much more jagged and distorted. You don't have to have a single high frequency signal to get high frequency components in the music. Line up a few lower frequency signals with the same phase, and the leading edge becomes very steep, the same as a single high frequency sine wave would be. Does this transient matter? I know it doesn't matter to you, but then I'm wondering how you managed to wander down into the "Audiophile" forum :-) It matters to me. So does the fact that the music 72 dB below full scale is horribly distorted. Like I say, this is a low frequency signal. Guessing from the plot in 8.1d, it must have several percent distortion. As a general rule, 1% distortion is just visible in a trace like this if I remember right, so this must be much worse than that. So to claim that the dynamic range exceeds this level just doesn't work for me. Here's another thing to think about. Say you have a signal at 12kHz that is pretty large in amplitude. I pick that frequency because it is low enough in frequency to be heard by most everyone, but high enough that the harmonics of it are outside the range of the anti-aliasing filter. So let's say something in the recording process goes horribly wrong and this signal gets badly distorted in a way that produces mainly harmonic distortion. What would a "conventional" distortion analyzer show? Since all of the harmonics are filtered out by the anti-aliasing filters in the CD encoding process, the distortion analyzer would read zero. For those that don't know, distortion analyzers work by using a very sharp notch filter to remove the fundamental (in this case the 12 kHz sine wave) and then measuring the amplitude of whatever is left. In this case, once you remove the fundamental, nothing is left because all of the harmonically related distortion components are removed by the anti-aliasing filter. So the distortion analyzer would read zero. But does this mean the signal really has no distortion? I think that if you examined it in the time domain, you would see it was still distorted. The same could be true for an 8kHz signal if it were distorted in such a way as to cause only odd order harmonics. This would happen if the signal got severely clipped. The first odd harmonic at 24kHz would not make it through the anti-aliasing filter (which is a very good thing). But does that mean the distortion was removed? The technique that started this thread would still show the distortion, while a conventional harmonic distortion analyzer would not. Im not saying that CDs suck. I only meant to say that the proposed technique might show some interesting results if applied to the CD encoding process that our present suite of sine wave based tools are not capable of showing. Terry -- TerryS ------------------------------------------------------------------------ TerryS's Profile: http://forums.slimdevices.com/member.php?userid=40835 View this thread: http://forums.slimdevices.com/showthread.php?t=87175
_______________________________________________ audiophiles mailing list audiophiles@lists.slimdevices.com http://lists.slimdevices.com/mailman/listinfo/audiophiles