Sorry, Dan, I can't go along with this - theoretically, and practically, 16 bits are fine. I've done various types of fiddling over the years, in experiments of reducing number of effective bits used for encoding, and application of dithering - and it all works like it's supposed to. If you're hearing distortion artifacts in some situation then it's because the implementation is faulty, or the design of the decoding mechanism is flawed. Apart from the single, badly mastered opera recording I've mentioned a couple of times, I haven't heard bad sound because of digital format "deficiencies" intruding.
Last edited:
Dan, unless one does something silly with the (analogue) gain control the distortion harmonics are always going to be buried in the listening room background noise - the signal level may drop, but the distortion levels stay constant, with respect to that background noise ... you just won't be able to hear any artifacts.
Somebody here (RNM ?) stated that the Philips engineers responsible for CD standard opined that 16 bit was not actually good enough.
Of course technology and market forces determined the Red Book standard.
I do agree that 16 bit on a correctly sounding system sounds quite fine.
However......
A few months back I repaired a vinyl system for a friend and played a mint Helen Reddy vinyl that I had on hand, and also digital copy that I am well familiar with.
I had run up the system on a selection of usb digital and CD before he came to collect and all sounded good.
When he arrived I played the HR vinyl and I was floored compared to the digital version.
So much more presence/articulation/clarity that I had not heard in a long time due to my digital only PB for yonks.
I have a Dual 505 and selection of cartridges in the garage...when I get a roundtuit I will try some more listening comparisons to refresh/confirm my memories of what vinyl can really do.
The digital PB system that I have running currently is quite stunningly good, that said I am playing mostly pop/rock/jazz which by definition run at elevated average levels, and do not really reveal really low level non linearities in DAC output stages as to be expected when running orchestral music.
Dan.
My listening room is well damped and essentially silent....come 7:00/8:00 pm there are no traffic/suburban/neighbour noises/sounds to interfere.....seriously.
Dan.
Last edited:
Last night, my wife (who was not in the kitchen) and I watched a delightfully goofy demonstration on Penn & Teller's TV series, the name of which is automatically censored by the forum software. It had nothing to do with audio, it was about organic food. They set up a booth at a farmers' market and had a side by side blind comparison of so-called organic versus non-organic produce.
Scroll to 2:45.
https://www.youtube.com/watch?v=8Zqe4ZV9LDs
Language warning for the sensitive- they talk pretty much the way I do. The adjectives are amazingly familiar, as are the excuses.
"Which of these feels better to eat"
Many folks preference for organic has nothing to do with taste.
About a thousand basic texts on signal processing. This is first principles stuff.
Stunning, I guess he doesn't believe in standards. BTW I hope Richard realizes eventually that the total THD + noise is always > the single highest peak spurious tone. His current definition of DNR is far more lenient than (dare I say) the standard one.
Between home PC and lunchtime smart phone I might have missed you link.
Please refresh me.
Thanks, Dan.
Dan, I suggested a simple experiment. Mathematically generate an arbitrarily complex waveform (like a multi-tone) at 64bit precision and truncate and dither it down to say 24 bits. Then subtract the original, the residue is indistinguishable from random noise. There is no systematic error on any time scale, but you have lost 3dB SNR. Folks show typical sine wave FFT demos but the principle works for any waveform. Dither far predates digital audio and has a sound mathematical background.
Last edited:
Ok, thanks for that Scott.
I should do the experiments you suggest, and also repeat my previous experiments.
The pdf you quote gives levels of quantisation noise lower than -100dB, with differing spectral nature according to the dithering type.
What is interesting is that these different ditherings are noted as sounding different and ranked according to perceived loudness.
This paper reinforces my previous statements about differences in audibility of differing dither types (Mash, Sony, Cool Edit etc).
The fact that eons ago I picked up on these to me audible differences prior to my in depth knowledge of dithering theory reinforces the fact of audibility of -100dB noise/distortions.
IMHO this info actually reinforces the notion that better than 16 bit systems are required.
Dan.
I should do the experiments you suggest, and also repeat my previous experiments.
The pdf you quote gives levels of quantisation noise lower than -100dB, with differing spectral nature according to the dithering type.
What is interesting is that these different ditherings are noted as sounding different and ranked according to perceived loudness.
This paper reinforces my previous statements about differences in audibility of differing dither types (Mash, Sony, Cool Edit etc).
The fact that eons ago I picked up on these to me audible differences prior to my in depth knowledge of dithering theory reinforces the fact of audibility of -100dB noise/distortions.
IMHO this info actually reinforces the notion that better than 16 bit systems are required.
Dan.
I hope you didn't miss this... "After quantization, the noise samples were amplified
by 40 dB for easy listening."
- Status
- Not open for further replies.