I thought you already have a conclusion of 40k bandwith, hence FS should be > 80k.
I originally chose 40KHz, because that is the practical limit of usable microphones, super-tweeters, and analog 30ips tape recorders. Going further is too difficult, without serious compromise in other areas. Now, we know that electronics can be 100KHz or even more, and this is important, because we can't have everything rolling off at once, or we get significantly less overall frequency response.
For more than the last 40 years we have known about the obvious problems with anti-aliasing filters, but we know that they are necessary with digital processing with finite sampling rates. How to build a better, more transparent filter has been the goal of many design engineers for more than 50 years.
A little history, over 50 years ago, Ampex used 50KHz clock for their first attempt at digital recording. I was there. A few years later, 1974, Phillips made their first digital recorder and they also used 50KHz. Why? Because that was the LOWEST Clock Rate even rationally practical to make an anti-aliasing filter pass 20KHz. Ampex and Phillips did not have to agree on this, they knew it as practical engineers.
Now where did 44KHz come from? It came from Sony. They insisted on it and we have suffered its consequences ever since. Every trick in the engineer's handbook has been used to 'improve' CD and a practical limit has been reached. However, who needs CD when we have DVD, SACD, Blue Ray, or digital streaming that can sample at much higher rates? Even the disc size is the same as CD.
It is just the stubbornness of many who have extensive CD collections, etc, that keeps us from moving forward. Look, keep your CD's, I'll keep mine, but be open to the better digital that is available today.
This latest discussion has some merit, but we knew the 44KHz was too low, since CD was first introduced. This latest discussion changes little or nothing, so let's go forward.
For more than the last 40 years we have known about the obvious problems with anti-aliasing filters, but we know that they are necessary with digital processing with finite sampling rates. How to build a better, more transparent filter has been the goal of many design engineers for more than 50 years.
A little history, over 50 years ago, Ampex used 50KHz clock for their first attempt at digital recording. I was there. A few years later, 1974, Phillips made their first digital recorder and they also used 50KHz. Why? Because that was the LOWEST Clock Rate even rationally practical to make an anti-aliasing filter pass 20KHz. Ampex and Phillips did not have to agree on this, they knew it as practical engineers.
Now where did 44KHz come from? It came from Sony. They insisted on it and we have suffered its consequences ever since. Every trick in the engineer's handbook has been used to 'improve' CD and a practical limit has been reached. However, who needs CD when we have DVD, SACD, Blue Ray, or digital streaming that can sample at much higher rates? Even the disc size is the same as CD.
It is just the stubbornness of many who have extensive CD collections, etc, that keeps us from moving forward. Look, keep your CD's, I'll keep mine, but be open to the better digital that is available today.
This latest discussion has some merit, but we knew the 44KHz was too low, since CD was first introduced. This latest discussion changes little or nothing, so let's go forward.
Last edited:
scottjoplin - the Quadrophenia recording suggestion was only an example of something with a lot of cymbal crash transients to listen to. Other than that, no relevance to the discussion on how anti aliasing filters effect the sound of such. Perhaps another might want a source with lots of "it" to listen to.
They also sound fine to me, but nothing like Keith Moon actually playing his drumset in my livingroom. Perhaps with a better source media, a millistep closer.
They also sound fine to me, but nothing like Keith Moon actually playing his drumset in my livingroom. Perhaps with a better source media, a millistep closer.
There is a difference between intensity and transparency in cymbal reproduction.
Last edited:
I think you will find this can be attributed to (at least) the bispectrum being the foundation of our aural perception, rather than our hearing extending beyond 20kHz and our brains exceeding the Uncertainty Principle (or at least its equivalent in the time-frequency plane).
The ear and brain functioning as a (third order) bispectral analyser does not preclude the useful application of more easily represented, conventional (second order) spectral analyses, however, just that we might consider a dynamic trade-off between time and frequency resolution somewhat commensurate with the convergence evident in our perceptual apparatus.
In such analyses, there might even be found measures that identify oft-reported differences between transient and steady-state audible phenomena. Ohm's Acoustical Law (that the ear is insensitive to phase), for example, might only be an approximation, albeit a very useful one in most cases.
Richard mentioned 100kHz as a minimum BW or sampling frequency (can't remember which) for digital. Anyway, what was your impression when using super tweeters?
Wow, you have nicely packed all existing prejudices in one posting.
I hope you don’t mind that I do no feel being a member of your “we knew etc” group.
Hans
I'm not sure what that means, spaciousness? And what about the source, what difference did that make?
Did you know about what Jn is talking about, or is that news to you and your contemporaries?
Did you know about what Jn is talking about, or is that news to you and your contemporaries?
Last edited:
Thought to mention that but didn't want to appear to be making a negative comment 😉
And where did the 44.1Khz came from and not an integer like 44 or 45 ?
Story goes that analogue NTSC video recorders at that time could record 525 interleaved lines of which 35 where reserved, giving 245 usable lines at 60Hz
So 245 lines at 60Hz, packed with three 16 bits samples per line, gives a sampling rate of 60 × 245 × 3 = 44.1 kHz.
Only point is that NTSC did not work with 30Hz per full frame but with 29.97Hz.
Hans
Story goes that analogue NTSC video recorders at that time could record 525 interleaved lines of which 35 where reserved, giving 245 usable lines at 60Hz
So 245 lines at 60Hz, packed with three 16 bits samples per line, gives a sampling rate of 60 × 245 × 3 = 44.1 kHz.
Only point is that NTSC did not work with 30Hz per full frame but with 29.97Hz.
Hans
What do you mean by timing? How do electronics change timing? ( timing and phase are not the same.)
