| Topic: |
Science > Physics |
| User: |
"Stijn Goris" |
| Date: |
18 Aug 2004 05:28:33 AM |
| Object: |
human hearing |
Hi all,
The human ear can hear from 20 to 20 000 hz. DAT system and the new sound
DVD have a bandwith of 48 000 hz. this gives 24 000 hz (Nyquist). Why do
people tell me that the clasic CD doesn't do because it only has a spectrum
of 22 000 Hz which even surpasses the human ear?
Regards
Sijn
.
|
|
| User: "" |
|
| Title: Re: human hearing |
18 Aug 2004 08:46:12 AM |
|
|
"Stijn Goris" <mepisto@hotmail.com> wrote in message news:<41232f38$0$315$ba620e4c@news.skynet.be>...
Hi all,
The human ear can hear from 20 to 20 000 hz. DAT system and the new sound
DVD have a bandwith of 48 000 hz. this gives 24 000 hz (Nyquist). Why do
people tell me that the clasic CD doesn't do because it only has a spectrum
of 22 000 Hz which even surpasses the human ear?
Digital can sample at that rate. That does not mean that it can
deliver 22KHz of signal. Consider, for example, that music is
rarely a single pure tone.
Recovering the signal from the digital sample is also an issue.
A simple amplifier is not sufficient. There needs to be some
interpolation of the digital signal into something more closely
resembling the signal that was originally sampled. Early digital
devices were poor at performing this function. This led to a
certain bad reputation among audiophiles.
I suspect there was also a certain amount of sour grapes on the
part of people who had spent many thousands of dollars on expensive
turntables, vinyl records, and various bits and bobs to go with.
Socks
.
|
|
|
| User: "Richard Herring" |
|
| Title: Re: human hearing |
18 Aug 2004 09:22:46 AM |
|
|
In message <c7976c46.0408180546.3d241bed@posting.google.com>,
puppet_sock@hotmail.com writes
"Stijn Goris" <mepisto@hotmail.com> wrote in message
news:<41232f38$0$315$ba620e4c@news.skynet.be>...
Hi all,
The human ear can hear from 20 to 20 000 hz. DAT system and the new sound
DVD have a bandwith of 48 000 hz. this gives 24 000 hz (Nyquist). Why do
people tell me that the clasic CD doesn't do because it only has a spectrum
of 22 000 Hz which even surpasses the human ear?
Digital can sample at that rate. That does not mean that it can
deliver 22KHz of signal.
Was Nyquist wrong, then?
Consider, for example, that music is
rarely a single pure tone.
Makes no difference. It's a linear superposition of them.
Recovering the signal from the digital sample is also an issue.
A simple amplifier is not sufficient. There needs to be some
interpolation of the digital signal into something more closely
resembling the signal that was originally sampled.
Something like a low-pass filter?
Early digital
devices were poor at performing this function.
Early A/D and D/A converters were _nonlinear_. It's hard to get
linearity over a dynamic range of <mumble> dB.
This led to a
certain bad reputation among audiophiles.
I suspect there was also a certain amount of sour grapes on the
part of people who had spent many thousands of dollars on expensive
turntables, vinyl records, and various bits and bobs to go with.
--
Richard Herring
.
|
|
|
| User: "Coolie" |
|
| Title: Re: human hearing |
18 Aug 2004 10:19:45 AM |
|
|
Look closer at Nyquist Theorem, it is not a hard limit all the time. (Same
thing with Shannon's Theorem, which is an approximation) Also look at the
required signal to noise ratio Vs sample rate and dynamic range. (this gets
you over to the Allen Variance )
"Richard Herring" <junk@[127.0.0.1]> wrote in message
news:mwDhwsl2Y2IBFwZ8@baesystems.com...
In message <c7976c46.0408180546.3d241bed@posting.google.com>,
puppet_sock@hotmail.com writes
"Stijn Goris" <mepisto@hotmail.com> wrote in message
news:<41232f38$0$315$ba620e4c@news.skynet.be>...
Hi all,
The human ear can hear from 20 to 20 000 hz. DAT system and the new
sound
DVD have a bandwith of 48 000 hz. this gives 24 000 hz (Nyquist). Why
do
people tell me that the clasic CD doesn't do because it only has a
spectrum
of 22 000 Hz which even surpasses the human ear?
Digital can sample at that rate. That does not mean that it can
deliver 22KHz of signal.
Was Nyquist wrong, then?
Consider, for example, that music is
rarely a single pure tone.
Makes no difference. It's a linear superposition of them.
Recovering the signal from the digital sample is also an issue.
A simple amplifier is not sufficient. There needs to be some
interpolation of the digital signal into something more closely
resembling the signal that was originally sampled.
Something like a low-pass filter?
Early digital
devices were poor at performing this function.
Early A/D and D/A converters were _nonlinear_. It's hard to get
linearity over a dynamic range of <mumble> dB.
This led to a
certain bad reputation among audiophiles.
I suspect there was also a certain amount of sour grapes on the
part of people who had spent many thousands of dollars on expensive
turntables, vinyl records, and various bits and bobs to go with.
--
Richard Herring
.
|
|
|
| User: "Richard Herring" |
|
| Title: Re: human hearing |
18 Aug 2004 11:00:25 AM |
|
|
In message <2ohaj8Fab2bdU1@uni-berlin.de>, Coolie <invalid@invalid.com>
writes
[please don't top-post - order corrected]
"Richard Herring" <junk@[127.0.0.1]> wrote in message
news:mwDhwsl2Y2IBFwZ8@baesystems.com...
In message <c7976c46.0408180546.3d241bed@posting.google.com>,
puppet_sock@hotmail.com writes
"Stijn Goris" <mepisto@hotmail.com> wrote in message
news:<41232f38$0$315$ba620e4c@news.skynet.be>...
Hi all,
The human ear can hear from 20 to 20 000 hz. DAT system and the new
sound
DVD have a bandwith of 48 000 hz. this gives 24 000 hz (Nyquist). Why
do
people tell me that the clasic CD doesn't do because it only has a
spectrum
of 22 000 Hz which even surpasses the human ear?
Digital can sample at that rate. That does not mean that it can
deliver 22KHz of signal.
Was Nyquist wrong, then?
Look closer at Nyquist Theorem, it is not a hard limit all the time.
Come on, you can do better than that. _Where_ is it not a "hard limit"?
(Same
thing with Shannon's Theorem, which is an approximation)
No, it's an existence proof for a limit.
Also look at the
required signal to noise ratio Vs sample rate and dynamic range.
That *is* Shannon's theorem.
(this gets
you over to the Allen Variance )
And that's about temporal variation in the sampling intervals.
--
Richard Herring
.
|
|
|
| User: "Ron Hardin" |
|
| Title: Re: human hearing |
18 Aug 2004 02:01:23 PM |
|
|
Richard Herring wrote:
Look closer at Nyquist Theorem, it is not a hard limit all the time.
Come on, you can do better than that. _Where_ is it not a "hard limit"?
I think there's prolate spheroidal wave functions that are a bandlimited
family and add to converge to any L2 integrable function over a finite time
interval. In short you can have as many wiggles as you want and still be
bandlimited at the family's width.
The interval winds up being in an exponentially small tail of something
really big, though.
That's not a sampling matter but just something about there being no wiggle
limit on a bandlimited function.
Unrelated :
It's not necessarily true that the ear does a fourier decomposition;
in fact it's very unlikely. Consider that the pitch of a sung note
is does not have to appear in the fourier decomposition. The pitch is
a modulation frequency.
--
Ron Hardin
rhhardin@mindspring.com
On the internet, nobody knows you're a jerk.
.
|
|
|
|
|
|
|
|
| User: "Randy Poe" |
|
| Title: Re: human hearing |
18 Aug 2004 09:26:34 AM |
|
|
"Stijn Goris" <mepisto@hotmail.com> wrote in message news:<41232f38$0$315$ba620e4c@news.skynet.be>...
Hi all,
The human ear can hear from 20 to 20 000 hz. DAT system and the new sound
DVD have a bandwith of 48 000 hz. this gives 24 000 hz (Nyquist). Why do
people tell me that the clasic CD doesn't do because it only has a spectrum
of 22 000 Hz which even surpasses the human ear?
I'm so glad you didn't crosspost to an audiophile newsgroup.
You've stepped into a religious war. Oversampling in a DAT
is fine, it gives you more capability to do signal
processing such as noise reduction. But as far as sound
reproduction, there is no reason to reproduce sounds
beyond 22 kHz. I think it's also silly to try to get
much beyond 16 bits accuracy. Each bit corresponds to
6 dB of power, so 16 bits = 96 dB dynamic range. What
is the point of 32-bit sound = 192 dB dynamic range?
Anybody who claims to be able to hear the differences
between 22 kHz and 48 kHz, or between 16 and 32 bits,
is (a) trying to sell you overpriced equipment, or
(b) trying to justify their purchase of overpriced
equipment.
People really are religious about this topic. Let the
flame wars begin.
- Randy
.
|
|
|
|
| User: "Ian Stirling" |
|
| Title: Re: human hearing |
18 Aug 2004 07:27:15 AM |
|
|
Stijn Goris <mepisto@hotmail.com> wrote:
Hi all,
The human ear can hear from 20 to 20 000 hz. DAT system and the new sound
DVD have a bandwith of 48 000 hz. this gives 24 000 hz (Nyquist). Why do
people tell me that the clasic CD doesn't do because it only has a spectrum
of 22 000 Hz which even surpasses the human ear?
Because they have golden ears.
You should promptly kill them and resell the ears.
.
|
|
|
|
|
Related Articles |
|
|
