The Logic Off Topic list

From: =?iso-8859-1?q?Pev?= <no_such_user@...>
Date: Thu Feb 14, 2002 0:30pm
Subject: Re: [GEN] Is 24 bits for A/D much more important than going up 
to 96k?

 > > However, still as a
 > > consequence of physics, the only sound at 24 khz that is
 > > reproducable with 48 khz equipment is a sine wave.
 > Square wave.
Yes and no.

Yes : at full rate, the bit-on bit-off behaviour will produce a square wave
in the *digital data*.

No : In practice this is not the case! All D->A converters will have a low
pass filter with a cut off at the nyquist frequency ; 24k in this case. This
will turn the square wave into a sine.

If you have a square wave at 24kHz with a 48kHz sample rate, you have some
serious aliasing problems!

HTH,

~Pev

From: "Markus Kaarlonen" <markus.kaarlonen@...>
Date: Thu Feb 14, 2002 1:01pm
Subject: Re: [GEN] 24 bits vs. 96K

> I disagree with this. Bit depth has nothing to do with dynamic range. Bit
> depth has to do with resolution. A 16 bit converter and 24 bit converter
> will clip at exactly the same point (provided they are accurately
> calibrated). The difference is the quality of the quiter bits, providing
> a more accurate depiction of the instrument being recorded.

Um, isn't this a bit like like saying "100 is not bigger than 10. It's just
ten times more accurate." ?

"Dynamic range" is the difference between the softest and loudest sound
producable in a certain system, right? In a digital system, the loudest
possible sound is very easy to determine, it's right at the point where all
bits are used and the sound is about to clip.

The less bits you use, the more quantization noise you get. When the sound
gets softer, you use less bits, and at some point the quantization noise
gets so loud that this can be considered the softest sound of this system.

Now, naturally more bits means a greater difference between the softest and
loudest sound, which equals more dynamic range. So, "a better/bigger dynamic
range" and "a more accurate depiction of the instrument being recorded" are
in essence the very same thing.

That's how I understand it, correct me if I'm wrong. :)

Markus "Captain" Kaarlonen
Lead Web Developer - www.madonion.com
captain@m...

From: Thomas Peitzmann <peitzmann@...-muenster.de>
Date: Thu Feb 14, 2002 1:13pm
Subject: Re: OT: 96kHz and stuff...

At 10:58 Uhr +0000 14.02.2002, "Mark Lennox" <mark@e...> wrote:
 >I know most people cant hear above say 17kHz or so, and thus the argument
 >for 96kHz sampling rate seems ridiculous.
 >
 >However, you have to remember that instruments, voices, etc... produce
 >overtones, harmonics, whatever stretching far above the audible threshold.
 >If these tones interact to form beat frequencies the resultant frequency
 >could be within the audible range.
 >
 >While recording you may record the beat frequency but not the original 
tones
 >that produced it - does anyone think that this would matter? Is this why
 >people claim that 96+kHz smapling rate is much better? Is it all in their
 >heads :)

Superimposing two frequencies (sinusoidal) on a linear system (like
air at reasonable pressure) produces no new frequencies. You hear the
beat frequency from audible tones, but that's a perceptive phenomenon
(which doesn't say it is non-existent). If you look at the
corresponding frequency spectrum it will still be just the two
sinusoids you started out from. So you can't hear the beats if you
can't hear the frequencies.

It's a totally different story for non-linear (modulating) systems,
though, which part of the human auditory system probably is. Then
it's a question of whether the cutoff of the high frequencies occurs
before or after the non-linearity.

Best regards, Thomas

From: "Richmond, James (James) ** CTR **" <richmondj@...>
Date: Thu Feb 14, 2002 1:37pm
Subject: RE: [LUG] Re: OT: 96kHz and stuff...

> --- In logic-users@y..., "Mark Lennox" <mark@e...> wrote:
> > I know most people cant hear above say 17kHz or so, and thus
> the argument
> > for 96kHz sampling rate seems ridiculous.
> ----
> perhaps by any scientific and measurable method its
> impossible to show that people can hear above such
> frequencies. also, our perceptions of what is and what isnt is
> usually based on our conscious awareness, it may well be that
> in some unconscious way we can indeed have a sense of what
> it is to experience these higher frequencies. ive heard so many
> comments regarding this - "it feels more airy", "it has more
> transparency", "the sound seems to have more life". these are of
> course only anecdotal and subjective opinions, yet they're the
> real experiences of real people, some of whom are highly
> regarded creative people! so, although a 96k sampling rate can
> seem ridiculous, it may well not be :-)

Well interestingly there are some studies that report the following:

96khz, 24bit (as opposed to 44.1/16bit) results in:
·much less "busy signal" break up - very good quality
· better separation of reverberation/room acoustics from instrument
output
· bass better balanced
· percussion (particularly cymbals) better
· some stereo image formation

BUT going up to 192khz, 24bit results in:

· detailed comparisons not yet performed on enough systems, but well
liked by (classical) artists after sessions
· no "busy signal" break up
· very good separation of reverberation and room acoustics
· strong stereo image formation, no observations on width so far

The bass problem is "commented on by experienced listeners, and shows up
particularly well on multi-mic'd rock music - it
can enable 192 kS/s material to be identified in absence of the analogue
source material for
comparison. The bass response of the ADC and DAC units in question is
flat down to 1 Hz, so there
is no question of it being due to bass roll off there."

The report is from http://www.dcsltd.co.uk/papers/effects.pdf and weel
worth a read.
DCSLTD are responsible for the worlds first 24bit A/D converter.

I'd be very interested in tracking a rock band on a new HD system @
192khz and comparing it with one tracked at 96khz to hear this for
myself.

>  Is this why
> > people claim that 96+kHz smapling rate is much better? Is it all
> in their
> > heads :)
> ---
> yes! name me one single perception which ISNT in the head!

Extra-sensory-perception would be the one that comes to mind.

James Richmond

From: Garth Hemphill <ghemphill@...>
Date: Thu Feb 14, 2002 2:25pm
Subject: Re: [GEN] 24 bits vs. 96K

> The difference is the quality of the quiter bits, providing
> a more accurate depiction of the instrument being recorded.
>
> best,
> Jon

Which is an increase in dynamic range!

Sorry, I was trying to keep the description less technical, and more in
laymans terms, but the end result of going from 16 to 24 bits is being able
to better hear dynamic range, as opposed to frequency response.


Garth Hemphill
Principal, GLH Design, Inc.
Resident Sound Designer, ACT San Francisco
USA-829 Sound Designer #0011

From: Lennart =?iso-8859-1?Q?W=E5hlin?= <tca@...>
Date: Thu Feb 14, 2002 3:25pm
Subject: Re: [GEN] 24 bits vs. 96K

Message: 22
    Date: Wed, 13 Feb 2002 20:09:42 -0700
    From: Jonathan Christensen <jon@s...>
Subject: Re: [GEN]  24 bits vs. 96K

 >Bit Rate determines dynamic range.  You will definitely hear mor of 
the soft
 >sounds, and really loud sounds without compression when you use 24 bits
 >instead of 16.  It is certainly a much more audible difference.

Jonathan Christensen answered:
 >I disagree with this. Bit depth has nothing to do with dynamic range. Bit
 >depth has to do with resolution. A 16 bit converter and 24 bit converter
 >will clip at exactly the same point (provided they are accurately
 >calibrated). The difference is the quality of the quiter bits, providing
 >a more accurate depiction of the instrument being recorded.

Eh...?
You just described the higher dynamics of 24bit....that you disagreed upon.
16 bit gives a theoretical dynamic range of 96dB, 24bit gives a
theoretical dynamic range of
144dB. A bit depth of 100000bit will will clip at exactly the same
level as anything else.

 >(provided they are accurately  calibrated)

Dynamic range has nothing to do with how strong you can play.
It is how quite you can play without distortion.

Best
Lennart


--

         TAKING CARE of AUDIO(tm)
             tca@t...

From: Jim Offerman <j.offerman@...>
Date: Thu Feb 14, 2002 0:39pm
Subject: Re: [LUG] OT: 96kHz and stuff...

Mark,

> I know most people cant hear above say 17kHz or so, and thus the argument
> for 96kHz sampling rate seems ridiculous.

You are mixing up two *very* different things: 17kHz refers to the
highest frequency people can hear, whilst the 96kHz refers to the
frequency at which the sound is sampled by digital recorder.

I.e. a 96kHz digital recorder samples the frequency of the sound that
is being recorded 96.000 times per second. This has *nothing* to do
with the actual frequency range of the sound being recorded.

Jim Offerman

www.jimofferman.nl
www.mp3.com/jimofferman/

From: Jonathan Christensen <jon@...>
Date: Thu Feb 14, 2002 4:12pm
Subject: Re: [GEN] 24 bits vs. 96K

>>Bit Rate determines dynamic range.  You will definitely hear mor of the soft
>>sounds, and really loud sounds without compression when you use 24 bits
>>instead of 16.  It is certainly a much more audible difference.

>The misconception is that
> the more bits the louder the signal, but that isn't accurate, and
> isn't what dynamic range is referring to (in this case, the change
> in the noise floor is the change in dynamic range).

Hi Terry,

Points taken. But I still disagree with the statement that "24 bits will
let you hear the really loud sounds without compression". The benefit of
24bit vs. 16bit is really more a matter of footroom than headroom.

best,
Jon


-- config--
Apple G4 Dual 800, 1.5 gig of ram, Mac OS 9.2.2, RME Hammerfall DSP/
Digiface, MOTU 2408, MOTU MTP AV (USB),  Apogee WC Master, Spirit Digital
328 (digital mixer/controller), ATI Radeon AGP, ATI Rage128 (second monitor).

From: Jonathan Christensen <jon@...>
Date: Thu Feb 14, 2002 4:14pm
Subject: Re: [GEN] 24 bits vs. 96K

>>>Bit Rate determines dynamic range.  You will definitely hear mor of
the soft
>>>sounds, and really loud sounds without compression when you use 24 bits
>>>instead of 16.  It is certainly a much more audible difference.
>
>>The misconception is that
>> the more bits the louder the signal, but that isn't accurate, and
>> isn't what dynamic range is referring to (in this case, the change
>> in the noise floor is the change in dynamic range).
>
>Hi Terry,
>
>Points taken. But I still disagree with the statement that "24 bits will
>let you hear the really loud sounds without compression". The benefit of
>24bit vs. 16bit is really more a matter of footroom than headroom.
>
>best,
>Jon


Ooops . .sorry Dedric. Didn't mean to flip flop your first and last names.

Jon

>> I disagree with this. Bit depth has nothing to do with dynamic
> range. Bit
>> depth has to do with resolution. A 16 bit converter and 24 bit
>
> I can see why one would assume, but bit depth does determine
> dynamic range in the form of power level, and the level of the
> noise floor.  16 bits give you 96dB of range (0dB to -96dB), and
> 24 give you 144dB (0 to -144dB).  Since waveforms are positive
> and negative going, this is often represented as +/- 48dB (16
> bits), or +/-72dB (24 bits), but power is an absolute valueIf the
> two were not relative to a fixed value, then you would have to
> resample to change bit depth, instead of dither/truncate/round or
> pad (in the case of increased bit depth).  More processing would
> be involved to redefine, say 1 bit per 2dB instead of the usual
> 3dB.  The only thing that usually doesn't change is 0dBFS, which
> is usually represented by all 1's in each bit position (but in
> internal processing that could represent another number too to
> handle overs and carry-over summing for example, but the fixed
> bit output would be limited to 0dBFS for power level purposes).
> The extra bits (down to 1 bit) simply represent a quiter range.  If
> this were not so, then 16 bit converters and 24 bit converters
> would clip at different power levels.  The misconception is that
> the more bits the louder the signal, but that isn't accurate, and
> isn't what dynamic range is referring to (in this case, the change
> in the noise floor is the change in dynamic range).
>
> Sampling frequency determines actual resolution.  Put bit depth
> on the x axis of a graph (+/- [dynamic range]/2) and sampling on
> the y axis (time in seconds, ms, or us) and you will see how
> resolution changes with sampling frequency.
>
> Sorry for the long OT reply.  Now related to Logic, anyone with
> Logic 5 have a report on the effect of the new POW-r dithering?
>
> Regards,
> Dedric
>
> --- In logic-users@y..., Jonathan Christensen <jon@s...> wrote:
>>> Bit Rate determines dynamic range.  You will definitely hear
> mor of the soft
>>> sounds, and really loud sounds without compression when
> you use 24 bits
>>> instead of 16.  It is certainly a much more audible difference.
>>
>> I disagree with this. Bit depth has nothing to do with dynamic
> range. Bit
>> depth has to do with resolution. A 16 bit converter and 24 bit
> converter
>> will clip at exactly the same point (provided they are accurately
>> calibrated). The difference is the quality of the quiter bits,
> providing
>> a more accurate depiction of the instrument being recorded.
>>
>> best,
>> Jon
>>

From: Ask Kaereby <ask@...>
Date: Thu Feb 14, 2002 5:32pm
Subject: [OT] 24/96 - it's all in the filter!

The reason why 96 kHz sounds better than 44.1 has to do with the
anti-alias filter. With 44.1 any frequency above 22.05kHz will produce
alias artifacts (as described by Nyquist) so they must be removed by
means of a filter. If you want 20-20kHz response, you must construct a
filter that (ideally) will pass 20kHz at full level and dampen
22.05kHz below the threshold of hearing (-96dB should do for 16bit).
Thats a filter slope of ~ 1000dB/octave!!!  AFAIK all constructed
filters of this type compromise either frequency response or
anti-aliasing and possibly also phase integrity. For comparison, with
96kHz a filter starting at 20kHz only needs a slope of ~ 100dB/oct. It
is possible to construct such a filter with only neglectible impact,
if any, on the audio quality.


Ask Kaereby

www.audioengineer.com/ask

> From: "Mark Lennox" <mark@e...>
> Reply-To: logic-users@yahoogroups.com
> Date: Thu, 14 Feb 2002 08:06:09 -0000
> To: <logic-users@yahoogroups.com>
> Subject: [LUG]  OT: 96kHz and stuff...
>
> I know most people cant hear above say 17kHz or so, and thus the argument
> for 96kHz sampling rate seems ridiculous.
>
> However, you have to remember that instruments, voices, etc... produce
> overtones, harmonics, whatever stretching far above the audible threshold.
> If these tones interact to form beat frequencies the resultant frequency
> could be within the audible range.
>
> While recording you may record the beat frequency but not the original tones
> that produced it - does anyone think that this would matter? Is this why
> people claim that 96+kHz smapling rate is much better? Is it all in their
> heads :)
>
> Having never heard a digital system better than my own (not bragging here,
> just lamenting my lack of big buck studio experience :( ) - A tango24
> interface into a cheap 'HI-FI' amp and a set of NS10's (I'm sure most
> peoples experience of high-end audio..), I'm torn between the argument that
> what we have now is way beyond what was available to musicians all through
> history and the argument that we must always strive for more.
>
> I sometimes feel that we have neglected to fully exploit a 'plateau' of
> technology before we rush headlong to the next big thing. I sometimes wonder
> who is left to discover the best mix of front-end/convertor/monitor that
> brings out the best in a given recording situation. There are long
> established practices for recording to tape - while the majority of these
> will long be cherished and valid, what needs to be updated for the digital
> medium?
>
> --
> Mark Lennox
> Technical Consultant
> ENDUSER
> Suite 40
> Guinness Enterprise Centre
> Taylors Lane
> Dublin 8
> Ireland
> Tel: +353 1 4100 665
> Fax: +353 1 4100 985
> web: http://www.enduser.com
> --

From: Hendrik Jan Veenstra <h@...>
Date: Thu Feb 14, 2002 3:06pm
Subject: Re: [LUG] [GEN] Is 24 bits for A/D much more important than 
going up to 96k?

Thoughts from the mind of monsdrum, 13-02-2002:

 >However, still as a consequence of physics, the only sound at 24 khz
 >that is reproducable with 48 khz equipment is a sine wave.

Shouldn't that be a square?  2 samples for an entire wave is just a
single 0 and 1 -- i.e. a square.  There's no way 2 samples can catch
the "flow" of a sine.

--
Hendrik Jan Veenstra  <h@k...>
Omega Art: http://www.ision.nl/users/h/index.html

From: Sean McCoy <osr@...>
Date: Thu Feb 14, 2002 8:00pm
Subject: Re: [LUG] [GEN] Is 24 bits for A/D much more important than going up to 96k?

>I have 24 bit Swissonics so
>it´s not like I´m saying not to buy it´s just that the improvement is not
>like the one from 8 to 16 bit, it´s much more subtle.
>Henrik

I don't find the difference between 16 and 24 bits to be subtle at
all.  It's certainly not as dramatic as the difference between 8 and 16,
but for me it's the sonic equivalent of "good" vs. "great." And I felt that
way with my first 24 bit experience, which was moving from the mediocre 16
bit converters in a CardD plus to the mediocre 24 bit converters in a
Frontier Tango 24.

Sean McCoy
Oregon Sound Recording

From: Steve Chandler <stevec@...>
Date: Thu Feb 14, 2002 8:31pm
Subject: Re: [GEN] 24 bits vs. 96K

> Jon wrote
>
> I disagree with this. Bit depth has nothing to do with dynamic range. Bit
> depth has to do with resolution. A 16 bit converter and 24 bit converter
> will clip at exactly the same point (provided they are accurately
> calibrated). The difference is the quality of the quiter bits, providing
> a more accurate depiction of the instrument being recorded.
>

At 6 dB per bit, bit depth is about nothing but dynamic range. Dynamic range after
all is nothing but the range between clipping and the noise floor. So at 16 bits
that range is 96 dB and at 24 bits it's 144 dB. That's the theoretical maximum
dynamic range. Add in the self noise from all your electronics in the chain
including A/D converters and what you get will be less dynamic range. I agree that
at 24 bits with good electronics the quality of the quieter bits will be improved.

Regarding 96K sampling rates, it's been demonstrated that many people can hear the
difference between a sine wave and a square wave at 15 kHz (these demonstrations
were done at numerous AES shows by Rupert Neve). Quite obviously some people can
hear the effect of a 45 kHz harmonic. A 44.1 kHz sampling rate can't indicate
anything about the shape of such a wave, but a 96 Khz sampling rate can.

Steve

From: "David Sane" <ready-fire-aim@...>
Date: Fri Feb 15, 2002 3:52am
Subject: Re: [LUG] Re: [GEN] 24 bits vs. 96K

 >
 > Sample rate determines frequency response.  At 44.1, the CD standard, you
 > get 20hz - 20Khz, and that about covers the range of human hearing.  What
 > you get at 96K sample rate is much higher freq. Response, and some of the
 > upper harmonics of a sound that you don't hear alone, that affect and
change
 > the sounds you do hear at lower sample rates.  This explains why 
people do
 > hear a difference between 96K and 44.1K.  It sounds "airier", more open
and
 > pristine.  It is hard to explain.  You need to do a direct comparison.
 >
 > Bit Rate determines dynamic range.  You will definitely hear mor of the
soft
 > sounds, and really loud sounds without compression when you use 24 bits
 > instead of 16.  It is certainly a much more audible difference.
 >

also, the use of higher sampling rates can improve phase coherency,
sometimes even dramatically.  there have been blind tests conducted (i
forgot the url extacly, sorry), in which 10 pros (using neve capricorn
consoles, earthworks mics, cranesong stc8 compressor) listened to a 16/44.1,
24/96, and 32/192 sampling rate recording of a string quartet.  all 10 were
blown away by the clarity, and separation of 32/192.  subjective, i know,
but still relevant (imho).

i would wonder though, how detrimental SRC would be on such a high quality
file?  i assume that the effects of aliasing/artifacts would scale
logrithmically.. i could be wrong=)

for now, i think i'll stick to 24/48.

ds

From: "Ronald C.F. Antony" <rcfa+logicaudio@...>
Date: Fri Feb 15, 2002 5:01pm
Subject: Re: [LUG] Re: [GEN] 24 bits vs. 96K

 >> Bit Rate determines dynamic range.  You will definitely hear mor of
 >> the soft
 >> sounds, and really loud sounds without compression when you use 24 bits
 >> instead of 16.  It is certainly a much more audible difference.
 >
 > I disagree with this. Bit depth has nothing to do with dynamic range.
 > Bit
 > depth has to do with resolution. A 16 bit converter and 24 bit converter
 > will clip at exactly the same point (provided they are accurately
 > calibrated). The difference is the quality of the quiter bits, providing
 > a more accurate depiction of the instrument being recorded.

Sounds like you have a different definition of "dynamic range" than
everyone else I know. More resolution in the Y-axis is the same
as "dynamic range". The point being, that you can *expand* a signal
until you have the same crappy definition for quieter bits as in a
standard 16-bit recording, and at that point 0dB will be a much
higher absolute SPL.

What is "accurately calibrated" is open to discussion. Since we do
not record absolute SPLs (otherwise we'd not have gain faders on
our preamps, and neither pads on the microphones, etc.) there is
now "absolute" standard for calibration.

Thus you can, depending on your equipment and gain settings, record
in 24-bit such that the least significant 16-bits end up with the
exact same values as a 16-bit recording, and what you get is 8-bit
worth of headroom, which corresponds to a factor of 256 on linear
scale, which I think corresponds to 24dB headroom, i.e. expansion
of the dynamic range.

To illustrate this in the graphical domain: if you go from a monitor
with a resolution of 640*480 pixels to one of 1280*960, you either
quadruple the resolution of the same image, or you can fit four times
as much information at the same resolution as before. It is the
combination dpi (dots per inch) setting and screen size which determine
which of the two (or what combination) is the case. The audio
equivalent to this are the SPL and gain settings in relation to the
bit depth.

Resolution and dynamic range are the flip sides of the same medal.

Ronald
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
~~~~
"The reasonable man adapts himself to the world; the unreasonable one
persists
in trying to adapt the world to himself. Therefore all progress depends
on the
unreasonable man."  G.B. Shaw  |  rcfa @ cubiculum . com  |  NeXT-mail
welcome

From: "Ronald C.F. Antony" <rcfa+logicaudio@...>
Date: Fri Feb 15, 2002 5:08pm
Subject: Re: [LUG] [GEN] Is 24 bits for A/D much more important than going up to 96k?

>> However, still as a
>> consequence of physics, the only sound at 24 khz that is
>> reproducable with 48 khz equipment is a sine wave.
>
> Square wave.

This depends on the filter. Might as well be sawtooth (if you assume
linear interpolation between sample points) or whatever else, rather
than hard level stepping.

The idea is however that any decent DA converter
filters the not explicitly sampled high-frequencies. So what you're
left with, ideally, after the filter is a sine wave. Otherwise
you'd have a stepped or edgy output signal even for higher
frequencies.

This, besides the effects of jitter, explain why there is such a
huge difference in sound between digital playback devices. If it
weren't for the difference in quality of the clock and filter
design, they should all sound the same, but they don't.

Ronald
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
~~~~
"The reasonable man adapts himself to the world; the unreasonable one
persists
in trying to adapt the world to himself. Therefore all progress depends
on the
unreasonable man."  G.B. Shaw  |  rcfa @ cubiculum . com  |  NeXT-mail
welcome