|
Home
/ FAQ
/ News Classifieds / Events |
Audio Asylum Thread Printer |
Get a view of an entire thread on one page |
68.3.132.52
| '); } else { document.writeln(''); } } else { document.writeln(''); } } else { document.writeln(''); } } // End --> |
In Reply to: RE: Your words, not mine. posted by theaudiohobby on November 02, 2009 at 14:17:27
"I do not agree, preferring a rolled-off (read equalised response) ala Ayre and the Wavelength is a matter of preference and subjective judgement rather than an objective qualitative assessment."
I've mentioned why the "rolloff" exists several times. It's not a typical FR rolloff like what exists in band-limited analog products. It's due to the beat/modulation, caused in most part by a sample rate that is too low. Since the signal modulates, peaks are sampled might be close to flat or maybe even lower than what the FR curve would suggest. That curve only suggest the "average" output at HF. But subjectively, the HF does not seem "rolled off", relative to a digital filtered signal that's flat to 20 kHz....... Since the time response is improved, the HF even seems greater in some cases.
When it comes to digital audio playback, the measured "HF rolloff" only tells half the story.
... for those DACs sounding better, nevermind how is it SOLELY responsible for them sounding better.
Awkward attempt at obfuscation, nothing more.
"I've mentioned why the "rolloff" exists several times. It's not a typical FR rolloff like what exists in band-limited analog products. It's due to the beat/modulation, caused in most part by a sample rate that is too low. Since the signal modulates, peaks are sampled might be close to flat or maybe even lower than what the FR curve would suggest. That curve only suggest the "average" output at HF. But subjectively, the HF does not seem "rolled off", relative to a digital filtered signal that's flat to 20 kHz....... Since the time response is improved, the HF even seems greater in some cases."Your comments are somewhat long-winded and per Werner post (and many others besides) not particularly correct.
Secondly, per my original post all the filters employed by Ayre has deeper roll-off than the alternative implementations. The LP filter -1.7dB@20Hz is fully 1.5dB off the Wolfson and Benchmark implementation, MP filter is worse as it's fully 3.5dB off the Wolfson implementation, Those are dramatic differences in any sense.
And no time response is not improved with MP filters as it sacrifices phase linearity for amplitude linearity and that makes the deep roll-off even harder to explain. The other factor that's supposedly improved is pre-ringing (in decent implementations) but that's at the expense of increased post-ringing.
Music making the painting, recording it the photograph
Edits: 11/03/09
I don't consider myself any kind of expert in filter design, but I do believe that the minimum phase / linear phase tradeoff dimension is independent of the high frequency roll off dimension. The duration of significant impulse response will depend heavily on the steepness of the cut off slope, and if aliasing/imaging is to be avoided this implies there is a tradeoff between impulse response and high frequency amplitude response. The phase response is a separate issue. One can shift the needed transient response to be after the impulse (minimum phase) or centered on the impulse (linear phase). One can even, for amusement, shift the needed transient response to entirely precede the impulse (maximum phase).
One can see these different dimensions with some file based SRCs, such as iZotope 64 bit SRC where both dimensions are exposed separately via the user interface. If one wants to experiment with maximum phase effects one can use minimum phase filters in conjunction with an audio editor that allows time reversal of files.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
You know a lot more about filter design than me your post address the issues more lucidly than I can manage. I certainly agree that high frequency roll-off is independent of the minimum phase/linear divide which was essentially one of the main points I attempted to put across in the previous post. Thanks for jogging my memory on the trade-off between impulse response and high frequency amplitude response.
The only issue I have is wrt time response which is what I believe Todd mentioned as opposed to transient response. Time response and phase response are obviously related terms and an essential difference between the Linear phase and minimum phase is that one exhibits constant group delay and the other doesn't, therefore by definition the time response of both filter is very different.
Music making the painting, recording it the photograph
Post a Followup: