Not knowing anything useful about the details of quantized D/A reconstruction, I'd naively assumed that JN's search was really about the effect of the shape of the "hold" on reconstruction timing. As in: does the inevitable RC-analog-circuit-at-this-point droop of reconstructed sample voltage, as opposed to idealized flat-top, shift the sample's effective timing? And if so, enough to matter?
All good fortune,
Chris
All good fortune,
Chris
I think it depends on the part. I don't know what ESS is doing, but the normal stuff like TI PCM17xx and standard Cirrus and AKM parts use cascaded linear phase FIR filters. I believe they are halfband types. Could be different for the newest stuff.
Are you joking? If you can not measure it, do you thing other can not, either?
I truly have no idea what you would like to discuss here about. If oversampling/decimation (downsampling + filtering) is good? IMO yes it is, for a number of good reasons (also a topic for discussion). Oversampling and filtering types/methods? Sure it can be discussed. SQ of oversampled vs. 16/44k? I am not the best to talk about. Sure as hell, I am not debating if the differences between 16/44k and an oversampled version are audible or not, neither do I know of any rigorous preference test for these, with (for example) the LPF type as a parameter, if you are aware of any, I am all ears. Given that the latest ADCs and DACs are implementing multiple filters, I may think that the selection is a matter of individual preference, rather than based on any performance metric.
May I suggest instead of sitting on the fence and grumbling to propose a clear topic for discussion and follow through with your results, perhaps supported by data? And as a matter of fact, what is your dog in this oversampling race?
Regarding the comment about the Maxwell daemon, I have to admit I’m almost instinctively sick of people claiming all over the internet that oversampling “brings a new dimension to the music”, “the soundstage is vastly improved” “the bass is deeper and the highs are more detailed”, all sort of things that somehow involve that the oversampling brings extra information to the original material. This being said, I may have wrongly anticipated such a story, for which I should apologize.
Last edited:
-3dB at 40Khz has been the analog minimum in audio since forever and I use that number as a minimum for accurate sound up to 20KHz. No good audio reason to change that number for CD/digital.
THx-RNMarsh
Last edited:
Of course, compromise is unavoidable. At least you can not stabilize your CFA amplifier. You should slow down the slew rate or reduce ULGF.
Yes, for audio purposes cascaded linear phase FIR filters are probably in all high end parts. I was thinking on a larger scale; for ADCs with oversampling in the hundreds of MHz (like required for wireless, for filtering out the quantization noise) such filters are too complex to design and implement in silicon, polyphase IIR filters are much more convenient (and, if memory serves, less power hungry).
Sure - but this article is about persons with severe hearing losses. So phase, as it would be discussed here, is not on the same planet as the article. Your argumentation is often quite sloppy in general. Many of us I suppose notice this - I mention it so that you may adjust as you see fit. You think you educate us but in reality, you rather embarrass yourself.
//
//
TNT not really,
RNM was a technical manager of the highest order. He is able to motivate folks to do cutting edge work. In case you haven't figured it out he is still doing that.
He knows what he hears and acts on it by doing much himself and accords respect to others who do similar things.
Curiousity is only a crime in limited places and to some, here.
RNM was a technical manager of the highest order. He is able to motivate folks to do cutting edge work. In case you haven't figured it out he is still doing that.
He knows what he hears and acts on it by doing much himself and accords respect to others who do similar things.
Curiousity is only a crime in limited places and to some, here.
I seem to recall a few years back, he got caught on his own forum lying (or at least pushing the acceptable boundaries of marketing) about the capabilities of one of his DAC boxes. I believe he was claiming it had a PLL system from one of his more expensive models but in reality it was just an AD1896 ASRC. Someone took extensive photos of the boards to prove it and an ugly exchange ensued on the forum.
I got the impression that it wasn't a sin of omission - that he did it intentionally to make it look like he'd brought more of the features down into the lower cost box, but only he knows for sure.
Overall the PDFs are free, seem correct, and arrive at the right conclusions though. They do lack context.
How many recordings have content above 20 kHz? A number of the tracks I bought off HDtracks @ 96 or 192k to test were clearly resampled 44.1 or 48k material and had no content above 20k. Some of the other tracks weren't resampled, but still had barely anything above 20k.
I have to confess, I don't listen to much classical, so maybe it's a bit different in that genre.
I was responding to your reply to syn08 about the audibility of group delay in speakers, usually measured in milliseconds, affecting mainly LF and hopefully at least very similar in both channels. ITD is measured in microseconds and is a measure of the difference (obviously) between the channels. But you knew all that, so I jokingly (hence the wink) hinted you were building a strawman argument. 🙄 🙂
On analog recordings, there is stuff ON SOME RECORDINGS up at 50 kHz.
See Tomlinson Holman et al.
But, how much of it, if any, folds down into the audio BW through IMD I could not tell you.
Seems it’s a moot point in digital though since most of it is brick wall filtered at 22 kHz and for the higher sample rates where it’s not it’s fiddled with anyway ( noise shaping etc)
Last edited:
- Status
- Not open for further replies.