Maybe a bit offtopic, but to update my knowledge bass, how do these get around the restriction that in a Fourier Transform is restricted from having any input below the Nyquist? Seems to me that you can't run a digital filter until it is digitized, and you can't digitize it until it is bandwidth limited.
And what about pre-ringing?
dave
I would expect that being so close, it would spew stuff down into the accepted human hearing range.
dave
Globulator, please do yourself a big favour: study and understand the sampling theorem and its proof as Shannon wrote it down. You might be in for a surprise or two.
And yes, what you wrote about phase and undersampling etc. is demonstrably wrong. Very wrong.
FT has nothing to do with this, but OK.
Your last sentence is correct. But DS ADCs run their front-end at several MHz. The Nyquist rate then is half of several MHz. The analogue filter in front of a DS modulator is first or second order, with a -3dB point at 100kHz or higher. This is entirely innocuous w.r.t. phase (and if it were not you can start throwing out your tweeters, cartridges, and tube amps as these are orders of magnitudes worse).
The initial sampling at MHz rates is followed with noise-shaped decimation and digital domain AA filtering. All linear phase.
There is. And at the ADC side it gets encoded into the recording. (Mind, pre-ringing at the DAC side is a total non-issue for any properly bandlimited input signal. Look up the theorem again!)
There is no doubt whatsoever that pre-ringing can be clearly audible when it happens in the audible band. This is a serious problem that plagues applications like subband coding (MP3) and digital speaker crossovers.
But once the ringing is moved outside of the range of hearing (as is the case with an AA filter for 44.1kHz: it rings at 22.05kHz) the situation is less clearly cut. Some claim audibility, but there is no formal proof. There is also no known mechanism in the cochlea that would support such audibility. I am considering doing an experiment, with blind listening panel, testing several AA filter types. But this would take a lot of time, which I haven't.
Thanks for the patronising, now please demonstrate what determines the phase and amplitude of a 22.05kHz signal in a 44.1kHz system.
If you can 😉
So my point stands.
At 22.05kHz the phase is set by the CD's 44.1kHz clock and the amplitude by the difference between the alternating values.
An externally hosted image should be here but it was not working when we last tested it.
There is no other interpretation, period. Why was this very very wrong before, but OK now?
No, at 22.05 (but not 22.04), you're beginning to violate the conditions of the sampling theorem (f < 0.5fs). That's the whole point of antialiasing filters. At 22.04, the sampling theorem proves (it's in any textbook on signal processing, and I don't think we're ready to throw out our MRI units and go back to X-rays) that the waveform is captured and reproduced exactly- no phase "anomalies."
DIY indeed. And what have you done lately, except for stubbornly refusing to learn and understand?
Here is a 22kHz sine sampled at 174.6kHz. A phase difference was introduced, the lower channel leading by 45 degrees.
The signal is 120 seconds long, in order to make it appear more or less band-limited later on.
An externally hosted image should be here but it was not working when we last tested it.
Here is what you get after downsampling the signal to 44.1kHz, using a decent anti-alias filter.
Shock! All phase information seems to have been killed.
An externally hosted image should be here but it was not working when we last tested it.
Don't worry. Zooming out of the same excerpt it seems that the amplitude of our sines is modulated. Weird, not? But these are just the samples, and not ready for consumption yet...
An externally hosted image should be here but it was not working when we last tested it.
Now we take the 44.1kHz signal and oversampling it again to 176.4kHz. The oversampling contains an anti-imaging filter, also known as a reconstruction filter. We need it to get back from the raw sampled data to the original band-limited signal.
And does the lower channel still lead the upper one by 45 degrees?
An externally hosted image should be here but it was not working when we last tested it.
To Globulator: of course I could have faked these images. There is no reason whatsoever to believe me or my sincerity.
Last edited:
OT personal posts removed. Stick to the topic. If you can not be civil, do not post in this thread.
So in other words they are not sampling at 44 Khz but much-much higher. Something i've been advocating since very early in this thread.
dave
Make us the same with a signal that is linearly sweeping from
20khz to 22khz and back to 20khz every 20 periods........
Would be curious to see the result..
Vinyl and tubes ....RIDICULOUS . maybe in 1940. Not in 2010 !
IT'S TIME FOR SERVERS , DAC , and super fast amplifiers. vinyl..., let me laugh !
IT'S TIME FOR SERVERS , DAC , and super fast amplifiers. vinyl..., let me laugh !
nirvana
You got most of that right. It's servers and wireless links etc. that degrade most of the sound quality potential obtainable by ripping from a quiet optimised PC. You need to get the .wav files directly to the DAC without additional processing.
Going this route results in SQ that is way ahead of even the majority of CD players , even those such as the AU$4,500 (originally) Marantz SA11 SACD player directly playing the CD into the same DAC.
But hey, what would I know, I am only a stupid old man !
However, quite a few others are now doing similar,
and it is so damn easy to demonstrate through high resolution gear into either speakers or headphones.
As for CD vs proper high resolution DVD-As of the same performance, there is simply no comparison.
However, I suspect that quite often the record companies may have resorted to upsampling with some of the older recordings to create the DVD-A. I can't really comment about SACD, because most affordable players are a joke, and not even capable of playing normal CD very well.
SACD can't normally be ripped to .wav files either, unlike DVD-A.
Yes , I can't prove my statements,other than by demonstration directly to others, as I have done on many occasions to several members of this forum,
and another forum. I am more than happy to do the same for any other Sydney DIYA members.
I also do regular comparison uploads for other people of some of my rips.
This is a quote from a current upload.
"To me one sounds very good, a lot breathiness and expression, with well defined instruments BUT the other sounds like she is up close and personal,
purring in your ears and the percussion sounds far more real.e.g. on the intro you have that lovely wooden "plock" sound that the other version doesn't quite deliver."
The track in question was "Do That To Me One More Time-Jheena Lodwick" from "Best Audiophile Voices 2"
BTW, the checkums were identical with both versions !!!
The better sounding one was ripped quite some time later using EAC,
and directly to a Corsair Voyager GT USB 2.0 pen.
To some of the VERY smart members participating in this thread, I would suggest listening with your ears, and not your test equipment,
unless of course you have test equipment that can reveal "Jitter". (for want of a more suitable word.)
I will now leave the academics in the thread to express their utter disbelief and heap tons of ridicule on me,as one of the moderating team did about 2 years ago, when the differences were less obvious due to further improvements,
mainly the result of the use of the space age 3M 2552 aluminium adhesive anti vibration tape on Optical writer, HDDs etc, and the disabling of PWM fan near the writer during ripping. With my Intel 965LTCK motherboard, the quiet aftermarket fan is much noisier than when run by direct DC.
I fitted a red LED to the front panel of the PC to remind me to turn the fan back on again after ripping was completed.
At normal operating temperatures, this red LED was pulsing at approximately 10 impulses per second.
I have experience in this area, as test equipment in the older Telephone Exchanges gave impulses of 10 IPS with 33 and 1/3 and 66 and 2/3 ratio.
SandyK
Last edited:
So, just to make sure that I am understanding you correctly -- Are you saying that you have two .wav files with the same checksum and one sounds better than the other?
And are you saying that transmitting a .wav file across a wireless link or storing it on a server HD will degrade it?
And are you saying that transmitting a .wav file across a wireless link or storing it on a server HD will degrade it?
Its not quite that straightforward. With Delta-Sigma the analog signal is sampled at some megahertz, but a complete digital sample is only produced and available at 44.1 kHz (or whatever the sampling frequency is configured to be).
YES.
Chris Connaker who is the founder of Computer Audiophile Forum, which has many prominent members such as the Chesky brothers, has also stated that "Spinning HDDs muddy the waters " A well controlled test at a Symposium Chris organised some time back, reported that there was almost universal agreement that SSD sounded better than HDD too.
SandyK
Yeah man! Right on! Lets serve up gigabytes of over compressed dreck through "super fast" multichannel chip amps. Woohoo! Break out the Red Bull! 😛
- Status
- Not open for further replies.