I posted theory? Where?
Is this your next question?
BTW, your question was,
and I answered more than once. Are you satisfied with it? I would like to know before I deal with your next question above.
You don't seem to have answered any of my questions actually, you probably think you have, but answering one question with another doesn't count, that's ok though, I didn't expect you too.
When someone says something sounds better they are making a subjective statement about their perception, it's not necessarily related directly to the fidelity of the reproduction, you just have to accept it.
When someone says something sounds better they are making a subjective statement about their perception, it's not necessarily related directly to the fidelity of the reproduction, you just have to accept it.
Last edited:
There is something else you need to accept, the nature of subjective impression. It's been proven to be all over the place (depends on personal mood, temperature, listening position, suggestions, preconceived notion...etc.) thus unreliable evaluation method. Only usefulness it has is personal amusement. I've already questioned the lack of reliability of such impressions on post #1823 and 1827.
Your inability to digest my answer isn't the fault of my answer. Did you watch and understand the content of workshop video I linked earlier? If not, watch or watch again.
As for the location of theory I posted...?
Last edited:
It was about your question on what sounding better is in this context which is sound replaying system.
good you chimed in and helped Evenharmonics - he was really loosing it, now things are back to top troll
I do, the notion that no one's listening impressions have any value to anyone else is truely laughable, may as well close down this site now and live in our own little worlds, like you.
Further to my posting #1829, I’m getting more and more convinced that testing Dac’s with a single pulse to show differences between various filter types, as several OS Dac’s have nowadays, are very suggestive but almost telling nothing at all.
Why do we see this ringing, that’s because we do a convolution in the time domain with the Fourier transform of the (complex conjugate) filter curve in the frequency domain.
For a brick wall filter, it’s time domain transform is a sinc, and there we have the pre and and post ringing when applying a convolution to a single pulse.
The alternative way would have been to transform this single pulse to the frequency domain, multiply it's spectrum with the brick wall filter curve, transform it back into the time domain, and getting of course exactly the same time signal with pre and post ringing.
Now let’s filter this just filtered signal again with the same filter.
It’s hard to image what the result of the convolution will be in the time domain, so let’s transform this signal again to the frequency domain.
What we get is again exactly the filter shape we just used multiplied with the transformed signal.
Now multiply the previous filter shape again with the spectrum, assuming a brick wall filter that’s flat for the in-band, and it will not change anything to the spectrum, so when going back to the time domain, we still get the same unchanged sinc signal that we had before.
So brick wall filtering a signal in a DAC that was already brick wall filtered at the recording stage, doesn’t change the signal and does not add any pre and post ringing !
So my question is: is it possible for a filter in the OS upsampling process to add any ringing to a signal that was 100% restricted in frequency to Fs/2 in the recording process ?
If the answer is no, all those different filters like halve band, apodizing, slow, fast etc do not make any sense as long as the content below Fs/2 minus transition band remains unchanged.
And since we know from NOS that when a somewhat less sharp OS filter that still leaves a small part of the first image directly above Fs/2, we won’t be able to hear it, so what ?
But the length, calculation precision of the OS filter plus dithering, quantizing and the avoidance of inter overshoots are very important as we have seen in this thread.
So when differences can be heard between different type of OS filters, this has IMO nothing to do with more or less pre or post ringing, because eventual pre and post ringing already happened before in the recording stage and can no longer be added in the upsampling process.
If true, perceived sound difference between OS filters must therefore have to do with the quality of the filter implementation and that a well implemented Sinc filter can’t probably be improved upon.
Hans
Why do we see this ringing, that’s because we do a convolution in the time domain with the Fourier transform of the (complex conjugate) filter curve in the frequency domain.
For a brick wall filter, it’s time domain transform is a sinc, and there we have the pre and and post ringing when applying a convolution to a single pulse.
The alternative way would have been to transform this single pulse to the frequency domain, multiply it's spectrum with the brick wall filter curve, transform it back into the time domain, and getting of course exactly the same time signal with pre and post ringing.
Now let’s filter this just filtered signal again with the same filter.
It’s hard to image what the result of the convolution will be in the time domain, so let’s transform this signal again to the frequency domain.
What we get is again exactly the filter shape we just used multiplied with the transformed signal.
Now multiply the previous filter shape again with the spectrum, assuming a brick wall filter that’s flat for the in-band, and it will not change anything to the spectrum, so when going back to the time domain, we still get the same unchanged sinc signal that we had before.
So brick wall filtering a signal in a DAC that was already brick wall filtered at the recording stage, doesn’t change the signal and does not add any pre and post ringing !
So my question is: is it possible for a filter in the OS upsampling process to add any ringing to a signal that was 100% restricted in frequency to Fs/2 in the recording process ?
If the answer is no, all those different filters like halve band, apodizing, slow, fast etc do not make any sense as long as the content below Fs/2 minus transition band remains unchanged.
And since we know from NOS that when a somewhat less sharp OS filter that still leaves a small part of the first image directly above Fs/2, we won’t be able to hear it, so what ?
But the length, calculation precision of the OS filter plus dithering, quantizing and the avoidance of inter overshoots are very important as we have seen in this thread.
So when differences can be heard between different type of OS filters, this has IMO nothing to do with more or less pre or post ringing, because eventual pre and post ringing already happened before in the recording stage and can no longer be added in the upsampling process.
If true, perceived sound difference between OS filters must therefore have to do with the quality of the filter implementation and that a well implemented Sinc filter can’t probably be improved upon.
Hans
Firstly, Hans, thanks for some content amongst all the noise recently.
Secondly - I think that Marcel put it this way - if you insist that the pre/post-ringing of the filter can be heard is the same as insisting that you can hear content in frequency above nyquist frequency.
So regarding the theory that this pre/post-ringing is inaudible, my gut feeling is that you're correct. Also thumbs up for everything you said in the last paragraph. I think that we don't have to rely on my gut feeling - somebody with more DSP and math knowledge might chime in...
Secondly - I think that Marcel put it this way - if you insist that the pre/post-ringing of the filter can be heard is the same as insisting that you can hear content in frequency above nyquist frequency.
So regarding the theory that this pre/post-ringing is inaudible, my gut feeling is that you're correct. Also thumbs up for everything you said in the last paragraph. I think that we don't have to rely on my gut feeling - somebody with more DSP and math knowledge might chime in...
I am not purposefully helping anybody, only speaking for myself.
So you think shared subjective listening impressions are useful to others, and everybody disagreeing is a troll, duly noted.
What if something in a reproduction system is reacting in a time-variant way to ultrasonic content? Could artifacts below 20kHz be created?
It shows what kind of filter is used in the DAC, but it is easy to misinterpret. In particular, it doesn't necessarily say much about the impulse response of the whole record-playback chain.
Yep.
Exactly, that's why they call it perfect reconstruction of a band-limited signal. I've made several attempts to get this same message across, but never managed.
Apodizing filters as defined by Peter Garde have a relatively smooth roll-off and a stopband that starts at the beginning of the transition band of the steep filters in the signal chain. That is, they have a narrower bandwidth than the other filters, so they can limit pre- and post-ringing at the expense of bandwidth.
Your cat or even a young human with an unusually wide hearing range may dislike it.
Besides, the ringing is ultrasonic anyway (if you are not a bat, cat, dog or young human with an unusually large hearing range).
The same reasoning holds for a well-implemented Parks-McClellan filter with a passband ripple much smaller than what was used in the pre- and post-echo test, doesn't it?
Subjective listening impressions are useful to those who trying sell things like boutique DACs, amps and cables.
Jazz Man, anyone can post anything on this forum as long as it's within established forum rules. You can post as many useless drivel about what sound is as you want. Others can post just as many to call yours out for what they are. Remember, you and dddac are free to ignore anyone you both want to.