Home |
Search |
Today's Posts |
#1
|
|||
|
|||
Gibbs Phenomenon_pinerton or others?
I will start out by saying this is way over my head-but I have been
told this is a signifcant problem that digital has that analog doesn't-and is audible- http://mwt.e-technik.uni-ulm.de/worl...ier/node4.php3 Comment. |
#2
|
|||
|
|||
The Gibbs phenomenon is simply what happens when higher frequency
components are removed from a Fourier reconstruction of a square wave. The resulting waveform is bandwidth limited and is analogous to brick-wall filtering. Providing that the removed components are beyond the frequency response of humans, this would not appear to be a signicant source of sound degradation nor would it be audible. randy wrote: I will start out by saying this is way over my head-but I have been told this is a signifcant problem that digital has that analog doesn't-and is audible- http://mwt.e-technik.uni-ulm.de/worl...ier/node4.php3 Comment. |
#3
|
|||
|
|||
In article , "jwvm"
wrote: The Gibbs phenomenon is simply what happens when higher frequency components are removed from a Fourier reconstruction of a square wave.... I thought it was when an otherwise fine vocal group decided to sing in falsetto and become disco damaged. Sorry.... |
#4
|
|||
|
|||
randy wrote:
I will start out by saying this is way over my head-but I have been told this is a signifcant problem that digital has that analog doesn't-and is audible- http://mwt.e-technik.uni-ulm.de/worl...ier/node4.php3 Comment. In audio there is *no* squarewave to be digitized, so this problem doesn't occurr. The maximal (theoretical) slew rate would be happening with a 20kHz signal at full output power and would already blow your tweeters, so do not try to create this signal. In a music feed the voltage will be down at least 12dB, which corresponds to max. power/16. Whoever told you this is technically as uninformed as you are. But this seems to be one of the prerequisites of high-end "gurus". -- ciao Ban Bordighera, Italy |
#5
|
|||
|
|||
"I will start out by saying this is way over my head-but I have been told
this is a signifcant problem that digital has that analog doesn't-and is audible-" It is the "audible" that poses the problem. As we have discussed here before, a digital recording of an lp was made and people were unable to distinguish between them using listening alone. What ever theoretical suggestion that something is audible must be established in practice. In doing so we can establish what are the thresholds for various artifacts in a signal, including this one. |
#6
|
|||
|
|||
On 27 Sep 2005 02:11:59 GMT, "randy" wrote:
I will start out by saying this is way over my head-but I have been told this is a signifcant problem that digital has that analog doesn't-and is audible- http://mwt.e-technik.uni-ulm.de/worl...ier/node4.php3 Comment. You have been told this by people with an agenda, who are trying to baffle you with science. The Gibbs phenomenon is simply what happens to a discontinuous function, such as a square wave, when you remove the higher frequency components, and this applies just as much to analogue as to digital. In any properly implemented digital sampling system, the *input* signal is band-limited to less than half the sampling rate, so that no aliasing takes place, and the output signal is a virtually perfect representation of that band-limited input signal. Analogue systems also have bandwidth limits, and a 5kHz squarewave fed through the ubiquitous 24/96 digital system will look much 'squarer' than the same signal fed through a standard 15ips studio analogue recorder. Note that a 20kHz square wave will come out of *both* machines as a sine wave - but it will be a *clean* sine wave out of the digital system. -- Stewart Pinkerton | Music is Art - Audio is Engineering |
#7
|
|||
|
|||
In article ,
"jwvm" wrote: The Gibbs phenomenon is simply what happens when higher frequency components are removed from a Fourier reconstruction of a square wave. The resulting waveform is bandwidth limited and is analogous to brick-wall filtering. Providing that the removed components are beyond the frequency response of humans, this would not appear to be a signicant source of sound degradation nor would it be audible. Not exactly. The Gibbs phenomenon is what happens to the partial sums of the Fourier series derived from a function with a jump discontinuity (such as a square wave). All partial sums exhibit the "overshoot", regardless of bandwidth. |
#8
|
|||
|
|||
Stewart Pinkerton wrote:
On 27 Sep 2005 02:11:59 GMT, "randy" wrote: I will start out by saying this is way over my head-but I have been told this is a signifcant problem that digital has that analog doesn't-and is audible- http://mwt.e-technik.uni-ulm.de/worl...ier/node4.php3 Comment. You have been told this by people with an agenda, who are trying to baffle you with science. The Gibbs phenomenon is simply what happens to a discontinuous function, such as a square wave, when you remove the higher frequency components, and this applies just as much to analogue as to digital. In any properly implemented digital sampling system, the *input* signal is band-limited to less than half the sampling rate, so that no aliasing takes place, and the output signal is a virtually perfect representation of that band-limited input signal. Analogue systems also have bandwidth limits, and a 5kHz squarewave fed through the ubiquitous 24/96 digital system will look much 'squarer' than the same signal fed through a standard 15ips studio analogue recorder. Note that a 20kHz square wave will come out of *both* machines as a sine wave - but it will be a *clean* sine wave out of the digital system. -- Stewart Pinkerton | Music is Art - Audio is Engineering I kind of wondered whether them trying to "baffle" me with science wasn't the case. |
#9
|
|||
|
|||
Stewart Pinkerton wrote:
On 27 Sep 2005 02:11:59 GMT, "randy" wrote: I will start out by saying this is way over my head-but I have been told this is a signifcant problem that digital has that analog doesn't-and is audible- http://mwt.e-technik.uni-ulm.de/worl...ier/node4.php3 Comment. You have been told this by people with an agenda, who are trying to baffle you with science. Well, if those people hear digital as a degraded signal, and the person they are explaining it to hears digital as a degraded signal, then the explanation is secondary to their experience. As long as no one is coerced into disbelieving their own ears, no harm done. On the other hand, bringing the focus over and over to the technical issues can easily fool people into thinking that what they hear is supposed to explained by the specs. Mike |
#10
|
|||
|
|||
Jenn wrote:
In article , "jwvm" wrote: The Gibbs phenomenon is simply what happens when higher frequency components are removed from a Fourier reconstruction of a square wave.... I thought it was when an otherwise fine vocal group decided to sing in falsetto and become disco damaged. Sorry.... LOL! thanks for bringing some humor into this ultra-serious, life-or-death discussion we're having here... Mike |
#11
|
|||
|
|||
On 29 Sep 2005 01:59:10 GMT, Billy Shears wrote:
In article , "jwvm" wrote: The Gibbs phenomenon is simply what happens when higher frequency components are removed from a Fourier reconstruction of a square wave. The resulting waveform is bandwidth limited and is analogous to brick-wall filtering. Providing that the removed components are beyond the frequency response of humans, this would not appear to be a signicant source of sound degradation nor would it be audible. Not exactly. The Gibbs phenomenon is what happens to the partial sums of the Fourier series derived from a function with a jump discontinuity (such as a square wave). All partial sums exhibit the "overshoot", regardless of bandwidth. However, the important thing is that such a discontinuous signal *never* reaches the A/D conversion point of a properly implemented digital audio system, hence the whole thing is a red herring from an audio point of view. -- Stewart Pinkerton | Music is Art - Audio is Engineering |
#12
|
|||
|
|||
I would be interested to hear of that study. I have conducted similar blind
A/B tests in my studio and without fail every listener can pick out the vinyl--based on the fact that it sounds much better. I used 96KHz 24bit sampling and playback directly through a high quality sampler--not via CD. Another interesting phenomena is that I've recorded a vinyl cut at both 44KHz and 88KHz (same system-24bit) and played them back to listeners. Everyone can hear the difference. But when I upsample the 44KHz to 88KHz there is a disagreement between sophisticated listeners and experienced muscians. The "listeners" like the upsampled sound better than native 88KHz sound, while musicians agree that the upsampled sound sounds better than 44KHz, but also state that it sounds wrong somehow! Another question for the digital experts--if all the music information is in 44KHz CD's, why is do muscians prefer the SACD for listening? wrote in message ... "I will start out by saying this is way over my head-but I have been told this is a signifcant problem that digital has that analog doesn't-and is audible-" It is the "audible" that poses the problem. As we have discussed here before, a digital recording of an lp was made and people were unable to distinguish between them using listening alone. What ever theoretical suggestion that something is audible must be established in practice. In doing so we can establish what are the thresholds for various artifacts in a signal, including this one. |
#14
|
|||
|
|||
vinylbigot wrote:
I would be interested to hear of that study. I have conducted similar blind A/B tests in my studio and without fail every listener can pick out the vinyl--based on the fact that it sounds much better. I used 96KHz 24bit sampling and playback directly through a high quality sampler--not via CD. Interesting. Please describe the setup in some more detail. Espcially as you call yourself 'vinylbigot', one would want to know about double-blinding, randomization procedures, level-matching,time-synching, and number of successes/trials. Another interesting phenomena is that I've recorded a vinyl cut at both 44KHz and 88KHz (same system-24bit) and played them back to listeners. Everyone can hear the difference. But when I upsample the 44KHz to 88KHz there is a disagreement between sophisticated listeners and experienced muscians. The "listeners" like the upsampled sound better than native 88KHz sound, while musicians agree that the upsampled sound sounds better than 44KHz, but also state that it sounds wrong somehow! Another question for the digital experts--if all the music information is in 44KHz CD's, why is do muscians prefer the SACD for listening? Assuming you know for sure that the SACD and CD have been mastered identically except for the final format, do they prefere it when they know it's SACD, or do they prefer it when they're presented as above? It's curious that the actual developers of DSD and DVD-A seem never to have published (or perhaps never even conducted) such trials. -- -S |
#15
|
|||
|
|||
On 30 Sep 2005 02:50:20 GMT, "vinylbigot"
wrote: I would be interested to hear of that study. I have conducted similar blind A/B tests in my studio and without fail every listener can pick out the vinyl--based on the fact that it sounds much better. I used 96KHz 24bit sampling and playback directly through a high quality sampler--not via CD. Not to trivialise anything, but I suggest that you are very clearly doing it wrong. The late, great, Gabe Weiner of PGM records used to do this all the time, and not one person could ever hear the insertion of an A/D-D/A chain into the signal - and that was at 16/44, not your 24/96. BTW, his studio main monitors were Wilson Grand SLAMMS, so we are not talking about some 'garage studio' here. Another interesting phenomena is that I've recorded a vinyl cut at both 44KHz and 88KHz (same system-24bit) and played them back to listeners. Everyone can hear the difference. Try it without telling them which is which. But when I upsample the 44KHz to 88KHz there is a disagreement between sophisticated listeners and experienced muscians. The "listeners" like the upsampled sound better than native 88KHz sound, Now y'see, that kinda *proves* that you're doing it wrong! while musicians agree that the upsampled sound sounds better than 44KHz, but also state that it sounds wrong somehow! Another question for the digital experts--if all the music information is in 44KHz CD's, why is do muscians prefer the SACD for listening? They don't. That statement is as true as yours, depending on the musicians you pick. wrote in message ... "I will start out by saying this is way over my head-but I have been told this is a signifcant problem that digital has that analog doesn't-and is audible-" It is the "audible" that poses the problem. As we have discussed here before, a digital recording of an lp was made and people were unable to distinguish between them using listening alone. What ever theoretical suggestion that something is audible must be established in practice. In doing so we can establish what are the thresholds for various artifacts in a signal, including this one. -- Stewart Pinkerton | Music is Art - Audio is Engineering |
#16
|
|||
|
|||
On 30 Sep 2005 02:42:43 GMT, wrote:
Stewart Pinkerton wrote: On 27 Sep 2005 02:11:59 GMT, "randy" wrote: I will start out by saying this is way over my head-but I have been told this is a signifcant problem that digital has that analog doesn't-and is audible- http://mwt.e-technik.uni-ulm.de/worl...ier/node4.php3 Comment. You have been told this by people with an agenda, who are trying to baffle you with science. Well, if those people hear digital as a degraded signal, and the person they are explaining it to hears digital as a degraded signal, then the explanation is secondary to their experience. As long as no one is coerced into disbelieving their own ears, no harm done. On the other hand, bringing the focus over and over to the technical issues can easily fool people into thinking that what they hear is supposed to explained by the specs. And the sort of technical nonsense being spouted about the Gibbs phenomenon can fool people into *thinking* that digital is a degraded signal. Expectation bias, remember? -- Stewart Pinkerton | Music is Art - Audio is Engineering |
#17
|
|||
|
|||
"Another question for the digital experts--if all the music information is
in 44KHz CD's, why is do muscians prefer the SACD for listening?" Because "muscians" are no different in the way the perception artifacts are produced which are not contained in the signal, if in fact this assumption about their listening habits is correct. They are as vulnerable to all the social contexts by which anyone comes to form their opinions, which might be quite different if testing them by listening alone to establish artifacts from signal. |
#18
|
|||
|
|||
Stewart Pinkerton wrote:
On 30 Sep 2005 02:42:43 GMT, wrote: Stewart Pinkerton wrote: On 27 Sep 2005 02:11:59 GMT, "randy" wrote: I will start out by saying this is way over my head-but I have been told this is a signifcant problem that digital has that analog doesn't-and is audible- http://mwt.e-technik.uni-ulm.de/worl...ier/node4.php3 Comment. You have been told this by people with an agenda, who are trying to baffle you with science. Well, if those people hear digital as a degraded signal, and the person they are explaining it to hears digital as a degraded signal, then the explanation is secondary to their experience. As long as no one is coerced into disbelieving their own ears, no harm done. On the other hand, bringing the focus over and over to the technical issues can easily fool people into thinking that what they hear is supposed to explained by the specs. And the sort of technical nonsense being spouted about the Gibbs phenomenon can fool people into *thinking* that digital is a degraded signal. Expectation bias, remember? Any technical comment made in the context of "what you should hear" can create expectation bias. Works just the same with technical nonsense and technical sense. Mike |