Home » The PARIS Forums » PARIS: Main » cubase mixing levels
|
|
|
|
|
|
|
|
| Re: cubase mixing levels [message #74988 is a reply to message #74981] |
Fri, 27 October 2006 20:16   |
Martin Harrington
Messages: 560
Registered: September 2005
|
Senior Member |
|
|
Can't tell you anything technically about the Cubase mix bus, (I use
Nuendo),but I think it's basically the same, but it's a fallacy that if you
record at lower levels you are protecting the file from clipping.
What you are doing is not using all the "bits" available to you, and
therefore start introducing unwanted artifacts into the mix.
If the "bits" aren't there on the original recording, and the levelis
cosewuently low at the mix bus, no matter what you do, you can't get those
bits back and the resolution and "size" of your mix has to suffer.
I record as hot as I can, and use the channel faders to mix, usually never
moving the master fader, although having said that, my mixes for TV/ Doco
work are not quite as complicated as most decent size music mixes would be.
--
Martin Harrington
www.lendanear-sound.com
"John" <no@no.com> wrote in message news:45429eda@linux...
> Martin, so do you know anything about the Cubase mix bus? Do they maybe
> mean that on mixdown you pull the faders way back but still record hot?
> Just wondering how the Cubase mix bus behaves.
>
> Thanks
>
> Martin Harrington wrote:
>> That's not true, and dont let anyone tell you it is.
>> You still need to get all levels as optimised as possible, as we all did
>> with tape.
>> Otherwise you are not using all the bits available to you, and noise will
>> be the end result.
>> This is why the good/great engineers are what they are...they make sure
>> the levels are hot....just not to the stage of distortion.
>> it's a balancing act, but, hey, who said anything done properly is easy.
>>
|
|
|
|
| Re: cubase mixing levels [message #74989 is a reply to message #74988] |
Fri, 27 October 2006 21:03 |
AlexPlasko
Messages: 211
Registered: September 2006
|
Senior Member |
|
|
hi martin
I think what john is referring to is what started this thread. We are
trying to emulate the way paris handles files at mixdown, not at recording
files, at mixdown.
chuck said that paris automatically and transparently cuts channel levels
by -22db, and then adds it back automatically when it hits the submix bus
much the way analog consoles do.
what we were toying with is if was possible to emulate that *effect* with
other daws by cutting channel levels 22db and making it back up at the
output bus.
what we dont know is how cubase/nuendo mix bus handles the files.or exactly
how paris does it for that matter.
If we can duplicate the way paris handles the mix bus DJ can sleep nights
again .
"Martin Harrington" <lendan@bigpond.net.au> wrote in message
news:4542c966$1@linux...
> Can't tell you anything technically about the Cubase mix bus, (I use
> Nuendo),but I think it's basically the same, but it's a fallacy that if
> you record at lower levels you are protecting the file from clipping.
> What you are doing is not using all the "bits" available to you, and
> therefore start introducing unwanted artifacts into the mix.
> If the "bits" aren't there on the original recording, and the levelis
> cosewuently low at the mix bus, no matter what you do, you can't get those
> bits back and the resolution and "size" of your mix has to suffer.
> I record as hot as I can, and use the channel faders to mix, usually never
> moving the master fader, although having said that, my mixes for TV/ Doco
> work are not quite as complicated as most decent size music mixes would
> be.
> --
> Martin Harrington
> www.lendanear-sound.com
> "John" <no@no.com> wrote in message news:45429eda@linux...
>> Martin, so do you know anything about the Cubase mix bus? Do they maybe
>> mean that on mixdown you pull the faders way back but still record hot?
>> Just wondering how the Cubase mix bus behaves.
>>
>> Thanks
>>
>> Martin Harrington wrote:
>>> That's not true, and dont let anyone tell you it is.
>>> You still need to get all levels as optimised as possible, as we all did
>>> with tape.
>>> Otherwise you are not using all the bits available to you, and noise
>>> will be the end result.
>>> This is why the good/great engineers are what they are...they make sure
>>> the levels are hot....just not to the stage of distortion.
>>> it's a balancing act, but, hey, who said anything done properly is easy.
>>>
>
>
|
|
|
|
| Re: cubase mixing levels [message #74995 is a reply to message #74989] |
Fri, 27 October 2006 22:04 |
Dedric Terry
Messages: 788
Registered: June 2007
|
Senior Member |
|
|
I agree with Martin completely. To add my own general opinion on this long
running "sound of summing" debate, that is fast become urban legend:
I think it is time to start debunking some of the fables around digital
summing. Paris cutting levels and then adding the gain back at the master
does one and only one thing: pushes all tracks down by 22dB to make it
easier to sum them well below 0dBFS, and then make some of it back before
the master so you didn't know what happened. If you are mixing properly in
any native DAW, you will in effect do the same thing - lower track levels
such that the peak of the summation remains below 0dBFS.
In simple math terms:
1- Native DAW: 2+2=4
2- Paris: ((2-1)+(2-1)) + 2 = 4
With SX you adjust the gain yourself on the way in, if you like, or better
yet, set the levels for the mix at hand as needed, as you go.
So why do this in Paris? I am guessing Paris had to convert all audio to
24-bit if sent to the native cpu for native plugins in order to prevent
clipping before you even start dropping faders on the mix (unless they
somhow converted to 32-bit float on the EDS chip first, which isn't a 32-bit
float chip, so that seems unlikely). (No one would want to mix with most
faders at -40dB - it would "seem" wrong). 24-bits for any portion of
summing (not tracking) is a limitation esp. if tracked audio files are near
0dBFS to begin with - there is no where to go with gain addition, and only
subtraction to work with. You can add a lot more -22dB peak audio files to
a mix without clipping (with all faders at 0), where only 2 audio files
peaking just below 0dBFS will automatically clip. At least that seems to be
the reasoning behind it, but imho, only applicable to prevent a potential
problem in the DAW itself, not as prescribed digital audio practice.
Back to native DAWs: While this approach may seem somehow capable of
producing a different sound to a final mix, the only think you gain by
lowering tracks by 22dB from the start is lower bit resolution. This is
especially true if you happen to record your tracks at lower levels (e.g.
-10dB peaks, then your peaks are now at -32dB, which is 1/3 the resolution
of 16 bit audio - that isn't insignificant). So, if you combine the flawed
advice to record at -10 to -20dB, with the concept of lowering all tracks by
22dB before you start mixing, and you end up removing most of the resolution
we work hard for with quality mics, preamps and converters, and mixing as
much more noise and quantization error than necessary.
As Martin suggested, record just below, or comfortably below clipping (as
the source dictates) to maximize your use of bit resolution - e.g. keep
audible that ambience or depth of the recording that tails off down at those
lower levels, rather than mixing it with mic self-noise, etc.
John, what I think the reference you quoted is actually saying (or should be
at least) is just that 24-bit digital (with a quality front end and
converters) affords a higher signal to noise ratio than analog did, so
pushing record levels to widen the gap between peaks and the noise floor
isn't as critical. But since quite a few really great mics have a noise
floor of around -70 to -74dB, we are still putting noise into half of the
bits of that glorious 24-bit range.
In terms of signal vs. noise - why record less signal when you can record
more?
Imho, there are a lot of "famous" engineers out there spouting complete
technical rubbish out of lack of true knowledge, or passing along
conversational heresay and conjecture. Sadly, engineering is becoming more
about letting the gear dictate the recording process than the engineer, and
I believe that's what's wrong with music today. Too my people think xyz
piece of gear can make a hit, a vibe, or a certain sound just because
someone else did it, but few actually use their skills to create the sound
by knowing what combinations of gear will help get them there.
9/10 times it isn't about obtaining one piece of gear to get "that sound",
but understanding the 1000 different possibilities and knowing how to use
any of them.
The Nuendo and SX audio engines are identical. They also are identical in
summing to Sequoia/Samplitude, and probably Logic, Audition, and DP too
(I've tested Nuendo and Sequoia beyond the limits of normal recording to
verify this, so this comes from experience and listening, not speculation or
internet urban legend).
Regards,
Dedric
On 10/27/06 10:03 PM, in article 4542d488@linux, "alex plasko"
<alex.plasko@snet.net> wrote:
> hi martin
> I think what john is referring to is what started this thread. We are
> trying to emulate the way paris handles files at mixdown, not at recording
> files, at mixdown.
> chuck said that paris automatically and transparently cuts channel levels
> by -22db, and then adds it back automatically when it hits the submix bus
> much the way analog consoles do.
> what we were toying with is if was possible to emulate that *effect* with
> other daws by cutting channel levels 22db and making it back up at the
> output bus.
> what we dont know is how cubase/nuendo mix bus handles the files.or exactly
> how paris does it for that matter.
> If we can duplicate the way paris handles the mix bus DJ can sleep nights
> again .
> "Martin Harrington" <lendan@bigpond.net.au> wrote in message
> news:4542c966$1@linux...
>> Can't tell you anything technically about the Cubase mix bus, (I use
>> Nuendo),but I think it's basically the same, but it's a fallacy that if
>> you record at lower levels you are protecting the file from clipping.
>> What you are doing is not using all the "bits" available to you, and
>> therefore start introducing unwanted artifacts into the mix.
>> If the "bits" aren't there on the original recording, and the levelis
>> cosewuently low at the mix bus, no matter what you do, you can't get those
>> bits back and the resolution and "size" of your mix has to suffer.
>> I record as hot as I can, and use the channel faders to mix, usually never
>> moving the master fader, although having said that, my mixes for TV/ Doco
>> work are not quite as complicated as most decent size music mixes would
>> be.
>> --
>> Martin Harrington
>> www.lendanear-sound.com
>> "John" <no@no.com> wrote in message news:45429eda@linux...
>>> Martin, so do you know anything about the Cubase mix bus? Do they maybe
>>> mean that on mixdown you pull the faders way back but still record hot?
>>> Just wondering how the Cubase mix bus behaves.
>>>
>>> Thanks
>>>
>>> Martin Harrington wrote:
>>>> That's not true, and dont let anyone tell you it is.
>>>> You still need to get all levels as optimised as possible, as we all did
>>>> with tape.
>>>> Otherwise you are not using all the bits available to you, and noise
>>>> will be the end result.
>>>> This is why the good/great engineers are what they are...they make sure
>>>> the levels are hot....just not to the stage of distortion.
>>>> it's a balancing act, but, hey, who said anything done properly is easy.
>>>>
>>
>>
>
>
|
|
|
|
| Re: cubase mixing levels [message #74998 is a reply to message #74995] |
Sat, 28 October 2006 02:51 |
Ted Gerber
Messages: 705
Registered: January 2009
|
Senior Member |
|
|
HI Dedric
Thank you for your excellent posts and the time taken on this topic. I
appreciate it.
As far as specific math goes, there is one statement from the link John
referenced that contradicts your statement regarding resolution at
various levels
Your quote here:
" if you happen to record your tracks at lower levels (e.g.
>-10dB peaks, then your peaks are now at -32dB, which is 1/3 the resolution
>of 16 bit audio - that isn't insignificant). "
The other statement said that 24bit recording at -48db is equal to a full
range
16bit recording...
These 2 statements look like they're talking about the same thing (apples
to apples). If they are, how are they reconciled? If they're not and I'm
misunderstanding...
Ted
Dedric Terry <dterry@keyofd.net> wrote:
>I agree with Martin completely. To add my own general opinion on this long
>running "sound of summing" debate, that is fast become urban legend:
>
>I think it is time to start debunking some of the fables around digital
>summing. Paris cutting levels and then adding the gain back at the master
>does one and only one thing: pushes all tracks down by 22dB to make it
>easier to sum them well below 0dBFS, and then make some of it back before
>the master so you didn't know what happened. If you are mixing properly
in
>any native DAW, you will in effect do the same thing - lower track levels
>such that the peak of the summation remains below 0dBFS.
>
>In simple math terms:
>
>1- Native DAW: 2+2=4
>2- Paris: ((2-1)+(2-1)) + 2 = 4
>
>With SX you adjust the gain yourself on the way in, if you like, or better
>yet, set the levels for the mix at hand as needed, as you go.
>
>So why do this in Paris? I am guessing Paris had to convert all audio to
>24-bit if sent to the native cpu for native plugins in order to prevent
>clipping before you even start dropping faders on the mix (unless they
>somhow converted to 32-bit float on the EDS chip first, which isn't a 32-bit
>float chip, so that seems unlikely). (No one would want to mix with most
>faders at -40dB - it would "seem" wrong). 24-bits for any portion of
>summing (not tracking) is a limitation esp. if tracked audio files are near
>0dBFS to begin with - there is no where to go with gain addition, and only
>subtraction to work with. You can add a lot more -22dB peak audio files
to
>a mix without clipping (with all faders at 0), where only 2 audio files
>peaking just below 0dBFS will automatically clip. At least that seems to
be
>the reasoning behind it, but imho, only applicable to prevent a potential
>problem in the DAW itself, not as prescribed digital audio practice.
>
>Back to native DAWs: While this approach may seem somehow capable of
>producing a different sound to a final mix, the only think you gain by
>lowering tracks by 22dB from the start is lower bit resolution. This is
>especially true if you happen to record your tracks at lower levels (e.g.
>-10dB peaks, then your peaks are now at -32dB, which is 1/3 the resolution
>of 16 bit audio - that isn't insignificant). So, if you combine the flawed
>advice to record at -10 to -20dB, with the concept of lowering all tracks
by
>22dB before you start mixing, and you end up removing most of the resolution
>we work hard for with quality mics, preamps and converters, and mixing as
>much more noise and quantization error than necessary.
>
>As Martin suggested, record just below, or comfortably below clipping (as
>the source dictates) to maximize your use of bit resolution - e.g. keep
>audible that ambience or depth of the recording that tails off down at those
>lower levels, rather than mixing it with mic self-noise, etc.
>
>John, what I think the reference you quoted is actually saying (or should
be
>at least) is just that 24-bit digital (with a quality front end and
>converters) affords a higher signal to noise ratio than analog did, so
>pushing record levels to widen the gap between peaks and the noise floor
>isn't as critical. But since quite a few really great mics have a noise
>floor of around -70 to -74dB, we are still putting noise into half of the
>bits of that glorious 24-bit range.
>
>In terms of signal vs. noise - why record less signal when you can record
>more?
>
>Imho, there are a lot of "famous" engineers out there spouting complete
>technical rubbish out of lack of true knowledge, or passing along
>conversational heresay and conjecture. Sadly, engineering is becoming more
>about letting the gear dictate the recording process than the engineer,
and
>I believe that's what's wrong with music today. Too my people think xyz
>piece of gear can make a hit, a vibe, or a certain sound just because
>someone else did it, but few actually use their skills to create the sound
>by knowing what combinations of gear will help get them there.
>
>9/10 times it isn't about obtaining one piece of gear to get "that sound",
>but understanding the 1000 different possibilities and knowing how to use
>any of them.
>
>The Nuendo and SX audio engines are identical. They also are identical
in
>summing to Sequoia/Samplitude, and probably Logic, Audition, and DP too
>(I've tested Nuendo and Sequoia beyond the limits of normal recording to
>verify this, so this comes from experience and listening, not speculation
or
>internet urban legend).
>
>Regards,
>Dedric
>
>On 10/27/06 10:03 PM, in article 4542d488@linux, "alex plasko"
><alex.plasko@snet.net> wrote:
>
>> hi martin
>> I think what john is referring to is what started this thread. We are
>> trying to emulate the way paris handles files at mixdown, not at recording
>> files, at mixdown.
>> chuck said that paris automatically and transparently cuts channel levels
>> by -22db, and then adds it back automatically when it hits the submix
bus
>> much the way analog consoles do.
>> what we were toying with is if was possible to emulate that *effect*
with
>> other daws by cutting channel levels 22db and making it back up at the
>> output bus.
>> what we dont know is how cubase/nuendo mix bus handles the files.or exactly
>> how paris does it for that matter.
>> If we can duplicate the way paris handles the mix bus DJ can sleep nights
>> again .
>> "Martin Harrington" <lendan@bigpond.net.au> wrote in message
>> news:4542c966$1@linux...
>>> Can't tell you anything technically about the Cubase mix bus, (I use
>>> Nuendo),but I think it's basically the same, but it's a fallacy that
if
>>> you record at lower levels you are protecting the file from clipping.
>>> What you are doing is not using all the "bits" available to you, and
>>> therefore start introducing unwanted artifacts into the mix.
>>> If the "bits" aren't there on the original recording, and the levelis
>>> cosewuently low at the mix bus, no matter what you do, you can't get
those
>>> bits back and the resolution and "size" of your mix has to suffer.
>>> I record as hot as I can, and use the channel faders to mix, usually
never
>>> moving the master fader, although having said that, my mixes for TV/
Doco
>>> work are not quite as complicated as most decent size music mixes would
>>> be.
>>> --
>>> Martin Harrington
>>> www.lendanear-sound.com
>>> "John" <no@no.com> wrote in message news:45429eda@linux...
>>>> Martin, so do you know anything about the Cubase mix bus? Do they maybe
>>>> mean that on mixdown you pull the faders way back but still record hot?
>>>> Just wondering how the Cubase mix bus behaves.
>>>>
>>>> Thanks
>>>>
>>>> Martin Harrington wrote:
>>>>> That's not true, and dont let anyone tell you it is.
>>>>> You still need to get all levels as optimised as possible, as we all
did
>>>>> with tape.
>>>>> Otherwise you are not using all the bits available to you, and noise
>>>>> will be the end result.
>>>>> This is why the good/great engineers are what they are...they make
sure
>>>>> the levels are hot....just not to the stage of distortion.
>>>>> it's a balancing act, but, hey, who said anything done properly is
easy.
>>>>>
>>>
>>>
>>
>>
>
|
|
|
|
| Re: cubase mixing levels [message #75000 is a reply to message #74995] |
Sat, 28 October 2006 03:25 |
John [1]
Messages: 2229
Registered: September 2005
|
Senior Member |
|
|
great post! thanks
Dedric Terry wrote:
> I agree with Martin completely. To add my own general opinion on this long
> running "sound of summing" debate, that is fast become urban legend:
>
> I think it is time to start debunking some of the fables around digital
> summing. Paris cutting levels and then adding the gain back at the master
> does one and only one thing: pushes all tracks down by 22dB to make it
> easier to sum them well below 0dBFS, and then make some of it back before
> the master so you didn't know what happened. If you are mixing properly in
> any native DAW, you will in effect do the same thing - lower track levels
> such that the peak of the summation remains below 0dBFS.
>
> In simple math terms:
>
> 1- Native DAW: 2+2=4
> 2- Paris: ((2-1)+(2-1)) + 2 = 4
>
> With SX you adjust the gain yourself on the way in, if you like, or better
> yet, set the levels for the mix at hand as needed, as you go.
>
> So why do this in Paris? I am guessing Paris had to convert all audio to
> 24-bit if sent to the native cpu for native plugins in order to prevent
> clipping before you even start dropping faders on the mix (unless they
> somhow converted to 32-bit float on the EDS chip first, which isn't a 32-bit
> float chip, so that seems unlikely). (No one would want to mix with most
> faders at -40dB - it would "seem" wrong). 24-bits for any portion of
> summing (not tracking) is a limitation esp. if tracked audio files are near
> 0dBFS to begin with - there is no where to go with gain addition, and only
> subtraction to work with. You can add a lot more -22dB peak audio files to
> a mix without clipping (with all faders at 0), where only 2 audio files
> peaking just below 0dBFS will automatically clip. At least that seems to be
> the reasoning behind it, but imho, only applicable to prevent a potential
> problem in the DAW itself, not as prescribed digital audio practice.
>
> Back to native DAWs: While this approach may seem somehow capable of
> producing a different sound to a final mix, the only think you gain by
> lowering tracks by 22dB from the start is lower bit resolution. This is
> especially true if you happen to record your tracks at lower levels (e.g.
> -10dB peaks, then your peaks are now at -32dB, which is 1/3 the resolution
> of 16 bit audio - that isn't insignificant). So, if you combine the flawed
> advice to record at -10 to -20dB, with the concept of lowering all tracks by
> 22dB before you start mixing, and you end up removing most of the resolution
> we work hard for with quality mics, preamps and converters, and mixing as
> much more noise and quantization error than necessary.
>
> As Martin suggested, record just below, or comfortably below clipping (as
> the source dictates) to maximize your use of bit resolution - e.g. keep
> audible that ambience or depth of the recording that tails off down at those
> lower levels, rather than mixing it with mic self-noise, etc.
>
> John, what I think the reference you quoted is actually saying (or should be
> at least) is just that 24-bit digital (with a quality front end and
> converters) affords a higher signal to noise ratio than analog did, so
> pushing record levels to widen the gap between peaks and the noise floor
> isn't as critical. But since quite a few really great mics have a noise
> floor of around -70 to -74dB, we are still putting noise into half of the
> bits of that glorious 24-bit range.
>
> In terms of signal vs. noise - why record less signal when you can record
> more?
>
> Imho, there are a lot of "famous" engineers out there spouting complete
> technical rubbish out of lack of true knowledge, or passing along
> conversational heresay and conjecture. Sadly, engineering is becoming more
> about letting the gear dictate the recording process than the engineer, and
> I believe that's what's wrong with music today. Too my people think xyz
> piece of gear can make a hit, a vibe, or a certain sound just because
> someone else did it, but few actually use their skills to create the sound
> by knowing what combinations of gear will help get them there.
>
> 9/10 times it isn't about obtaining one piece of gear to get "that sound",
> but understanding the 1000 different possibilities and knowing how to use
> any of them.
>
> The Nuendo and SX audio engines are identical. They also are identical in
> summing to Sequoia/Samplitude, and probably Logic, Audition, and DP too
> (I've tested Nuendo and Sequoia beyond the limits of normal recording to
> verify this, so this comes from experience and listening, not speculation or
> internet urban legend).
>
> Regards,
> Dedric
>
> On 10/27/06 10:03 PM, in article 4542d488@linux, "alex plasko"
> <alex.plasko@snet.net> wrote:
>
>> hi martin
>> I think what john is referring to is what started this thread. We are
>> trying to emulate the way paris handles files at mixdown, not at recording
>> files, at mixdown.
>> chuck said that paris automatically and transparently cuts channel levels
>> by -22db, and then adds it back automatically when it hits the submix bus
>> much the way analog consoles do.
>> what we were toying with is if was possible to emulate that *effect* with
>> other daws by cutting channel levels 22db and making it back up at the
>> output bus.
>> what we dont know is how cubase/nuendo mix bus handles the files.or exactly
>> how paris does it for that matter.
>> If we can duplicate the way paris handles the mix bus DJ can sleep nights
>> again .
>> "Martin Harrington" <lendan@bigpond.net.au> wrote in message
>> news:4542c966$1@linux...
>>> Can't tell you anything technically about the Cubase mix bus, (I use
>>> Nuendo),but I think it's basically the same, but it's a fallacy that if
>>> you record at lower levels you are protecting the file from clipping.
>>> What you are doing is not using all the "bits" available to you, and
>>> therefore start introducing unwanted artifacts into the mix.
>>> If the "bits" aren't there on the original recording, and the levelis
>>> cosewuently low at the mix bus, no matter what you do, you can't get those
>>> bits back and the resolution and "size" of your mix has to suffer.
>>> I record as hot as I can, and use the channel faders to mix, usually never
>>> moving the master fader, although having said that, my mixes for TV/ Doco
>>> work are not quite as complicated as most decent size music mixes would
>>> be.
>>> --
>>> Martin Harrington
>>> www.lendanear-sound.com
>>> "John" <no@no.com> wrote in message news:45429eda@linux...
>>>> Martin, so do you know anything about the Cubase mix bus? Do they maybe
>>>> mean that on mixdown you pull the faders way back but still record hot?
>>>> Just wondering how the Cubase mix bus behaves.
>>>>
>>>> Thanks
>>>>
>>>> Martin Harrington wrote:
>>>>> That's not true, and dont let anyone tell you it is.
>>>>> You still need to get all levels as optimised as possible, as we all did
>>>>> with tape.
>>>>> Otherwise you are not using all the bits available to you, and noise
>>>>> will be the end result.
>>>>> This is why the good/great engineers are what they are...they make sure
>>>>> the levels are hot....just not to the stage of distortion.
>>>>> it's a balancing act, but, hey, who said anything done properly is easy.
>>>>>
>>>
>>
>
|
|
|
|
| Re: cubase mixing levels [message #75008 is a reply to message #74995] |
Sat, 28 October 2006 07:47 |
Neil
Messages: 1645
Registered: April 2006
|
Senior Member |
|
|
Dedric Terry <dterry@keyofd.net> wrote:
>I agree with Martin completely. To add my own general opinion
>on this long running "sound of summing" debate,
Great post, Dedric! BTW, "The Sound of Summing" - wasn't that a
song by Summon & Nullfunkel?
Anyway, a couple of things I wanted to bring up. You said:
>In simple math terms:
>1- Native DAW: 2+2=4
>2- Paris: ((2-1)+(2-1)) + 2 = 4
>With SX you adjust the gain yourself on the way in, if you like, or better
>yet, set the levels for the mix at hand as needed, as you go.
I just wanted to point out that if anyone thought I meant to
adjust your levels "on the way in" (i.e.: using the Channel
input trim control), when I had said to default all your
channels to -6, for example, as a starting point - that I did
NOT, in fact, mean the input trims!!! I meant the regular
ol' "please make it louder or softer" control. :D
>So why do this in Paris? I am guessing Paris had to convert all audio to
>24-bit if sent to the native cpu for native plugins in order to prevent
>clipping before you even start dropping faders on the mix (unless they
>somhow converted to 32-bit float on the EDS chip first, which isn't a 32-bit
>float chip, so that seems unlikely). (No one would want to mix with most
>faders at -40dB - it would "seem" wrong). 24-bits for any portion of
>summing (not tracking) is a limitation esp. if tracked audio files are near
>0dBFS to begin with - there is no where to go with gain addition, and only
>subtraction to work with. You can add a lot more -22dB peak audio files
to
>a mix without clipping (with all faders at 0), where only 2 audio files
>peaking just below 0dBFS will automatically clip. At least that seems to
be
>the reasoning behind it, but imho, only applicable to prevent a potential
>problem in the DAW itself, not as prescribed digital audio practice.
>
>Back to native DAWs: While this approach may seem somehow capable of
>producing a different sound to a final mix, the only think you gain by
>lowering tracks by 22dB from the start is lower bit resolution. This is
>especially true if you happen to record your tracks at lower levels (e.g.
>-10dB peaks, then your peaks are now at -32dB, which is 1/3 the resolution
>of 16 bit audio - that isn't insignificant). So, if you combine the flawed
>advice to record at -10 to -20dB, with the concept of lowering all tracks
by
>22dB before you start mixing, and you end up removing most of the resolution
>we work hard for with quality mics, preamps and converters, and mixing as
>much more noise and quantization error than necessary
Which is why I suggest starting at -6... maybe -10 if you're
going to be loading up a buttload of tracks. Think about it
with an analog analogy again, gang: If your analog mixing
console goes up to +10 on each channel, would you start a mix
with every fucking fader maxed out at +10??? Hit "play", and
how would THAT summing buss sound right about then? No, to
start out with, you'd bring all the faders down to 0, or -5,
or -10, or whatever the your comfortable starting point was,
knowing the console & how much headroom it's got, etc., right?
Pushing everything up to +10 to start off with would be just
simply way too much, yes?
So why are people not willing to get their heads around the
fact that in digital anything past "0" is "way too much"?
The only reason your DAW has +6 or +8 on any given channel is
in case you fuck up & record the cowbell track on your band's
cover version of "Mississippi Queen" at peak levels of -17db
.... "0" gain on the channel level just wouldn't cut it at that
stage, the cowbell just wouldn't be audible enough to drive
that tune. :D
So, just like in the analog console, where 40+ channels
recorded to needle-bending levels on overbiased tape machines,
and every channel set to +10 at the start of your mixdown
session would sound like crap; so does 40+ channels recorded to
nice hot levels, barely missing overs by a tenth of a db,
with every channel set to "0" result in a similar thing.
Neil
|
|
|
|
| Re: cubase mixing levels [message #75011 is a reply to message #74998] |
Sat, 28 October 2006 08:30 |
Dedric Terry
Messages: 788
Registered: June 2007
|
Senior Member |
|
|
Hi Ted,
16 bit and 24 bit both only represent up to 0dB full scale (FS). The
dynamic range afforded by 24-bit extends down to -144dB rather than the
-96dB of 16 bit. That's what we are interested in with digital audio, not
the theoretical limits above 0.
Once in the DAW, -48dB is represented by the lower 8 bits in 16-bit words,
and the lower 16 bits in 24-bit words. That's probably how they came to
that conclusion, but it's mathematically incorrect since the same bit
actually represents -48dB, you just add an extra 8 onto the bottom of the
dynamic range, not the top as the quote seems to assume.
Regards,
Dedric
On 10/28/06 3:51 AM, in article 4543283c$1@linux, "Ted Gerber"
<tedgerber@rogers.com> wrote:
> Your quote here:
>
> " if you happen to record your tracks at lower levels (e.g.
>> -10dB peaks, then your peaks are now at -32dB, which is 1/3 the resolution
>> of 16 bit audio - that isn't insignificant). "
>
> The other statement said that 24bit recording at -48db is equal to a full
> range
> 16bit recording...
>
> These 2 statements look like they're talking about the same thing (apples
> to apples). If they are, how are they reconciled? If they're not and I'm
> misunderstanding...
>
> Ted
|
|
|
|
| Re: cubase mixing levels [message #75017 is a reply to message #75011] |
Sat, 28 October 2006 09:43 |
audioguy_editout_
Messages: 249
Registered: December 2005
|
Senior Member |
|
|
This is a multi-part message in MIME format.
--------------090600040301020502030508
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
Ah, now, this is where people get confused.
The assumption that FULL bits (all 1111's) represents 0dbfs,
and that empty bits (all 00000's) represents the noise floor
is false.
The full bits actually represent the maximum positive
amplitude, and the empty bits represent the maximum
*negative* amplitude of the bi-phase audio signal. (I have
attached a pic below of a 4 bit signal capture, the vertical
axis shows the 4 bits, while the horizontal axis shows the
sample rate)
So, what is actually happening in 24 bit vs 16 bit is that
there are more bits to represent the vertical axis. This
means that "no signal" on the input of the recorder would
yield a number half way up, not all zeros.
Does this help?
David.
Dedric Terry wrote:
> Hi Ted,
>
> 16 bit and 24 bit both only represent up to 0dB full scale (FS). The
> dynamic range afforded by 24-bit extends down to -144dB rather than the
> -96dB of 16 bit. That's what we are interested in with digital audio, not
> the theoretical limits above 0.
>
> Once in the DAW, -48dB is represented by the lower 8 bits in 16-bit words,
> and the lower 16 bits in 24-bit words. That's probably how they came to
> that conclusion, but it's mathematically incorrect since the same bit
> actually represents -48dB, you just add an extra 8 onto the bottom of the
> dynamic range, not the top as the quote seems to assume.
>
> Regards,
> Dedric
>
> On 10/28/06 3:51 AM, in article 4543283c$1@linux, "Ted Gerber"
> <tedgerber@rogers.com> wrote:
>
>
>>Your quote here:
>>
>>" if you happen to record your tracks at lower levels (e.g.
>>
>>>-10dB peaks, then your peaks are now at -32dB, which is 1/3 the resolution
>>>of 16 bit audio - that isn't insignificant). "
>>
>>The other statement said that 24bit recording at -48db is equal to a full
>>range
>>16bit recording...
>>
>>These 2 statements look like they're talking about the same thing (apples
>>to apples). If they are, how are they reconciled? If they're not and I'm
>>misunderstanding...
>>
>>Ted
>
>
--------------090600040301020502030508
Content-Type: image/jpeg;
name="Pcm.jpg"
Content-Transfer-Encoding: base64
Content-Disposition: inline;
filename="Pcm.jpg"
/9j/4AAQSkZJRgABAQECWAJYAAD/2wBDAAUDBAQEAwUEBAQFBQUGBwwIBwcH Bw8LCwkMEQ8S
EhEPERETFhwXExQaFRERGCEYGh0dHx8fExciJCIeJBweHx7/2wBDAQUFBQcG Bw4ICA4eFBEU
Hh4eHh4eHh4eHh4eHh4eHh4eHh4eHh4eHh4eHh4eHh4eHh4eHh4eHh4eHh4e Hh4eHh7/wAAR
CAC8APoDASIAAhEBAxEB/8QAHwAAAQUBAQEBAQEAAAAAAAAAAAECAwQFBgcI CQoL/8QAtRAA
AgEDAwIEAwUFBAQAAAF9AQIDAAQRBRIhMUEGE1FhByJxFDKBkaEII0KxwRVS 0fAkM2JyggkK
FhcYGRolJicoKSo0NTY3ODk6Q0RFRkdISUpTVFVWV1hZWmNkZWZnaGlqc3R1 dnd4eXqDhIWG
h4iJipKTlJWWl5iZmqKjpKWmp6ipqrKztLW2t7i5usLDxMXGx8jJytLT1NXW 19jZ2uHi4+Tl
5ufo6erx8vP09fb3+Pn6/8QAHwEAAwEBAQEBAQEBAQAAAAAAAAECAwQFBgcI CQoL/8QAtREA
AgECBAQDBAcFBAQAAQJ3AAECAxEEBSExBhJBUQdhcRMiMoEIFEKRobHBCSMz UvAVYnLRChYk
NOEl8RcYGRomJygpKjU2Nzg5OkNERUZHSElKU1RVVldYWVpjZGVmZ2hpanN0 dXZ3eHl6goOE
hYaHiImKkpOUlZaXmJmaoqOkpaanqKmqsrO0tba3uLm6wsPExcbHyMnK0tPU 1dbX2Nna4uPk
5ebn6Onq8vP09fb3+Pn6/9oADAMBAAIRAxEAPwD6T13xP4gttI1q8sLS2lGl ao0M7iEuY7RY
VkaQRmRN7jd0DDI5AJ+U9xWLKPCwjub+RNI22t0WnmKx/upyACWbs5BUevIF bVAGe2oXYYga
HqDAHqHgwf8AyJSf2jef9ALUf++4P/jtaNFAGd/aN5/0AtR/77g/+O0f2jef 9ALUf++4P/jt
aNFAGd/aN5/0AtR/77g/+O0f2jef9ALUf++4P/jtaJIAJJAA6k1zyvceJiWh mlttDzgSRMUk
vvdWHKRe4wX6gheXmUradTalR57ybtFbv+t2+i/RNnN3/wAWLK38UX3h638I eKNQu7FkS5e1
jtWijZkDgFjOOx/PI6g1Nf8AxMeyg86TwD4ukX0hWzkb8luSa5bRLW2svin4 9tbO3it4I7mx
CRxIFVR9jj4AHArp6Vpdy/aUFooXXdvX5W0XzTGWvxSF1am6tvAvieaEZyyT 6ccEdQf9L4I9
OtdL4e8THXdA07W7DQ9UNnqFrFdQF2gVjHIgdcjzeDgjiuI1rTElSa9tJPsl 8IyPNUZEgA+7
Iv8AGP1HYit34E6lb3fwp8LWarJFcWuh2KvFKAGKGBNkgwSCrAZBHfIPIIAp NO0gnRi4udJ3
S3XVf5rz77pXV+o/tG8/6AWo/wDfcH/x2j+0bz/oBaj/AN9wf/Ha0aKs5jO/ tG8/6AWo/wDf
cH/x2j+0bz/oBaj/AN9wf/Ha0aKAM7+0bz/oBaj/AN9wf/HaP7RvP+gFqP8A 33B/8drRooAz
v7RvP+gFqP8A33B/8do/tG8/6AWo/wDfcH/x2tGigDO/tG8/6AWo/wDfcH/x 2j+0bz/oBaj/
AN9wf/Ha0aKAM7+0bz/oBaj/AN9wf/HaP7RvP+gFqP8A33B/8drRooAzv7Rv P+gFqP8A33B/
8do/tG8/6AWo/wDfcH/x2tGigDO/tG8/6AWo/wDfcH/x2j+0bz/oBaj/AN9w f/Ha0aKAM7+0
bz/oBaj/AN9wf/HaP7RvP+gFqP8A33B/8drRooAowX1zJMqPo99CpOC7vCVX 3OJCfyFXqKKA
OH13wTfarp+tWw1h7I31+91F5Kq64McaDeHQ8jYTx/e74FduoIUAsWIHJPeu E1vxFrMWh+JL
u2uAbjStRdIY4IUZ5IkijkK7XYbsBmJAIYgYGDXeUAZz6LZsxYzalknJxqNw B+Qek/sSy/57
an/4Mrj/AOLpXvdSDELosrAHg/aI+f1pPt2p/wDQEl/8CI/8aAD+xLL/AJ7a n/4Mrj/4uj+x
LL/ntqf/AIMrj/4uj7dqf/QEl/8AAiP/ABrk/iL451nw5Fp9nYeFprzUtVlk gtEN3Eqqyxs5
dsnkAL04z6ilJpK7Lp03Ukox6nLfEi0t9V+JOjeG4NX8QRWCabfyahDFqtys Nywa0AjYiQZK
rLkjBGJPXpaXwxZqoVdX8VgAYAHifUeP/I9ZNpN4hvfFGjXc/g+fTLTTNMvr d5JdRhmkuJri
a1cudvcmFyT6tXS/ab3/AKBr/wDf1P8AGpiravc1r1FK0IfCtvPu/V/grLoU dM8L6Tp1xd3F
s2pG4vHV7mabVLmaSVlUKu53kLHCgAc8Yq9/Zlt/z0vf/AyX/wCKo+03v/QN f/v6n+NH2m9/
6Br/APf1P8as5xG0u1ZSrSXhBGCPtkv/AMVXPy+B7G00+CPRL/X7SWzhWG0Q eIb4IsS7f3I/
fZRCFUfLjGB1xiuh+03v/QNf/v6n+NH2m9/6Br/9/U/xpSSkrM0pVJU5KUf6 /wCHOP1u2ji0
PT9U03W/FkMx1jToHSTxHfuV330MUsbq0xB4Z1I5r2z+xLL/AJ7an/4Mrj/4 uvEviDDrMdsl
3p2kGVZNRsJ54WuEQNJDdQyKwboCyx7CT32dADXXJ8Uddg1bSrLVPh9d2Uep XqWUUw1SCTbI
wYrkDnHympi29HuXXpxi1OHwvby7r5flZ9Tv/wCxLL/ntqf/AIMrj/4uj+xL L/ntqf8A4Mrj
/wCLo+3an/0BJf8AwIj/AMaPt2p/9ASX/wACI/8AGrMA/sSy/wCe2p/+DK4/ +Lo/sSy/57an
/wCDK4/+Lo+3an/0BJf/AAIj/wAaPt2p/wDQEl/8CI/8aAD+xLL/AJ7an/4M rj/4uj+xLL/n
tqf/AIMrj/4uj7dqf/QEl/8AAiP/ABo+3an/ANASX/wIj/xoAP7Esv8Antqf /gyuP/i6P7Es
v+e2p/8AgyuP/i6Pt2p/9ASX/wACI/8AGj7dqf8A0BJf/AiP/GgA/sSy/wCe 2p/+DK4/+Lo/
sSy/57an/wCDK4/+Lo+3an/0BJf/AAIj/wAaPt2p/wDQEl/8CI/8aAD+xLL/ AJ7an/4Mrj/4
uj+xLL/ntqf/AIMrj/4uj7dqf/QEl/8AAiP/ABo+3an/ANASX/wIj/xoAP7E sv8Antqf/gyu
P/i6P7Esv+e2p/8AgyuP/i6Pt2p/9ASX/wACI/8AGj7dqf8A0BJf/AiP/GgA /sSy/wCe2p/+
DK4/+Lo/sSy/57an/wCDK4/+Lo+3an/0BJf/AAIj/wAaPt2p/wDQEl/8CI/8 aAH2+k2sEyyp
LfllOQHv53X8QzkH8av1Qt7vUHmVJdJkiQnlzOh2/gDmr9AGBqOveGbG3uLy 7kiWO0umWeQW
rt5MgUFnbCnaArAmQ/KAeSK365K98L6s9rqtvaavpypqeom6nS5055UaIxon kkCZcg7ASc4I
JG3BrrVyAATk9zQAUVnPoWjuxZtOtyxOSSnU0n9gaL/0Dbb/AL4oA0q8b+IM smpePfBWtFib
V9Su7exXsY1s5t0v/A26diqIe5qT9orRtCk8MaLpDWscTajrtnCwiGGMfmDc foCVH1Iqtpnw
+8F6bfx39j4dsYbqMEJKFJZcjBxk9xUP3pW7HVH91R5usvy6/e9Pk11Oooql /ZWnf8+cP/fN
H9lad/z5w/8AfNWcpdoql/ZWnf8APnD/AN80f2Vp3/PnD/3zQBdoql/ZWnf8 +cP/AHzR/ZWn
f8+cP/fNAFi6ghuraW2uEEkMqFHU9GUjBFcbq08x1bwfYXbl7qz8UWsbu3WV CkuyT/gQ6/7Q
Ydq6r+ytO/584f8Avmue8Z+F/D12ltdalpFrdWcUgW6jkTgITxJ9UbBz2UvU T0946sP+8Tov
rt6/8Hb1t2Pb6K8B03wZ4X0X4u+DF0vRra1W6+3JOEziRRblgCM84IzXtn9g aL/0Dbb/AL4q
zlNKis3+wNF/6Btt/wB8Uf2Bov8A0Dbb/vigDSorN/sDRf8AoG23/fFH9gaL /wBA22/74oA0
qKzf7A0X/oG23/fFH9gaL/0Dbb/vigDSorN/sDRf+gbbf98Uf2Bov/QNtv8A vigDSorN/sDR
f+gbbf8AfFH9gaL/ANA22/74oA0qKzf7A0X/AKBtt/3xR/YGi/8AQNtv++KA NKis3+wNF/6B
tt/3xR/YGi/9A22/74oA0qKoW+jaVbzLNDYQRyIcqwXkGr9AHBah4h1KGw1W WPUhvtvFVlYR
nZHxBJNarJGeP7skvP3h68V3tYV94u0OzjuXnuZB9lmlhmHktlTFF5shxjJV U5yODwBkkA7i
srKGUhlIyCDwRQBnvNroYhNO00rngm/cEj6eTWD498U634U8LXOuSaFp12IZ IYhEupum5pZU
iXnyTgAuCeDwK67ev95fzrzj9o66t0+FtzavceVLeX1lDAV5O8XUT/yQmlJ8 quaUqbqzUF1/
q/yOP8Qy+L/G9zZ302i6FaQ2+o20sbjV5ZCsdvMWcIv2YbvMYD5srwq8HGT1 Xmar/wA+dl/4
Ft/8bqe1igtraK2gVY4okCIo6KoGAPyqXcvqPzpQVlqVXqKpNuO2y9Ft/wAH zKfmar/z52X/
AIFt/wDG6PM1X/nzsv8AwLb/AON1c3L6j86Ny+o/OqMSn5mq/wDPnZf+Bbf/ ABujzNV/587L
/wAC2/8AjdXNy+o/OjcvqPzoAp+Zqv8Az52X/gW3/wAbo8zVf+fOy/8AAtv/ AI3VzcvqPzo3
L6j86AKfmar/AM+dl/4Ft/8AG6ZN/aM0Lwy2Fi8bqVZWumIYHgg/u6v7l9R+ dG5fUfnQNNp3
Rwmk3GqJ8UPBdhJBbSXdhLfQjfcsPNX7LlWLbM/dI5wcsGHFe1+dr3/QN0z/ AMGD/wDxmvIt
YnsdO+MXgXUrmRIhJJd2zue+6EqgP/AnP/fRr2qO5tpSRHcROR1CuDUQ093s dGJSk1VW0tfn
1X36+jRS87Xv+gbpn/gwf/4zR52vf9A3TP8AwYP/APGa0d6f3l/Ojen95fzq zmM7zte/6Bum
f+DB/wD4zR52vf8AQN0z/wAGD/8AxmtHen95fzo3p/eX86AM7zte/wCgbpn/ AIMH/wDjNHna
9/0DdM/8GD//ABmtHen95fzo3p/eX86AM7zte/6Bumf+DB//AIzR52vf9A3T P/Bg/wD8ZrR3
p/eX86N6f3l/OgDO87Xv+gbpn/gwf/4zR52vf9A3TP8AwYP/APGa0d6f3l/O jen95fzoAzvO
17/oG6Z/4MH/APjNHna9/wBA3TP/AAYP/wDGa0d6f3l/Ojen95fzoAzvO17/ AKBumf8Agwf/
AOM0edr3/QN0z/wYP/8AGa0d6f3l/Ojen95fzoAo28ustMonsLCOIn5mS9dm A9gYhn8xV+kD
KTgMPzpaAOQ1LwNBfPfySahIkl5PcOXSMApHPAsLoMnrhFYN6gcYyK6yCKOC FIYlCxxqFUDs
AMAVynxRbxJHoyy+GpM3QSZEhRiskkzRMISuFIIWTaSGwuMljhSD1qbtg343 Y5x0zQBQbQ9F
ZizaPp7MTkk2yEk/lXlPxp0vTLjwrrWpW+nWkUNhd2FlbmOBV3SNe25mbgc/ wp7FZB3rufH/
AIg1bwr4O1jxFKumGOwtZJkVzJ8zAfIpx6nA/GvLL6D4la/4Kt/Dt2PCMNjN LbXE00bXDTsU
nSdm5G0szKc9ssah6ySOqn+7oyn1ei/X8LL0bOz/ALL0z/oHWf8A34X/AAo/ svTP+gdZ/wDf
hf8ACk/4mvpZfm1H/E19LL82qzlF/svTP+gdZ/8Afhf8KP7L0z/oHWf/AH4X /Ck/4mvpZfm1
H/E19LL82oAX+y9M/wCgdZ/9+F/wo/svTP8AoHWf/fhf8KT/AImvpZfm1H/E 19LL82oAX+y9
M/6B1n/34X/Cj+y9M/6B1n/34X/Ck/4mvpZfm1H/ABNfSy/NqAF/svTP+gdZ /wDfhf8ACj+y
9M/6B1n/AN+F/wAKT/ia+ll+bUf8TX0svzagCnrHhbw7q9o1tqGiadOhB2l7 ZCUJBG5SR8rY
PUVg/DXQfD8XxX0yCPw9o9tMvh/UI7xIbKONXljubIBiAO4JYZ7P711X/E19 LL82rk9St/Fu
kfEaw8SaB/YZnubG4sHivTL5RZjFJuGzncVtwPTCfSolo0zqo/vKcqfVar5b /hr8ke0/2Fof
/QG07/wFT/Cj+wtD/wCgNp3/AICp/hXE+AfE/jnVvEesaJrdp4cjlsLO0u0k s3m2ss73CYO7
uDb/APj1drnXv7umf99P/hVnKL/YWh/9AbTv/AVP8KP7C0P/AKA2nf8AgKn+ FJnXv7umf99P
/hRnXv7umf8AfT/4UAL/AGFof/QG07/wFT/Cj+wtD/6A2nf+Aqf4Umde/u6Z /wB9P/hRnXv7
umf99P8A4UAL/YWh/wDQG07/AMBU/wAKP7C0P/oDad/4Cp/hSZ17+7pn/fT/ AOFGde/u6Z/3
0/8AhQAv9haH/wBAbTv/AAFT/Cj+wtD/AOgNp3/gKn+FJnXv7umf99P/AIUZ 17+7pn/fT/4U
AL/YWh/9AbTv/AVP8KP7C0P/AKA2nf8AgKn+FJnXv7umf99P/hRnXv7umf8A fT/4UAL/AGFo
f/QG07/wFT/Cj+wtD/6A2nf+Aqf4Umde/u6Z/wB9P/hRnXv7umf99P8A4UAS QaPpEEyzQaXY
xSIcq6W6BgfYgVeqhbnWfOXzxYeVn5thfdj2zV+gDH8U+ItP8OW1vPqAlIuZ TDEsYBLOEZ9o
yRkkIQFHJOAATWxWJ4w8M6f4p05bDUnnWBSxxEV5LIyZ+YHBG7IYYIIBBFbd AHl3xu1PTdS8
IeJdPXULRo7HQL64kUTL887QSpEgGeSP3jEdQfLNR6VqOnjTLQG+tQRCmR5q /wB0e9avxA05
7X4NeOLm8ijW9vNI1C4mxg7SYHCLn/ZRUXI6lSe9UtKjj/su0+Rf9Qnb/ZFR BaXfU6MTJcyh
HaKt/m/vvbysL/aWnf8AP/a/9/l/xo/tLTv+f+1/7/L/AI1Y8uP/AJ5r+VHl x/8APNfyqznK
/wDaWnf8/wDa/wDf5f8AGj+0tO/5/wC1/wC/y/41Y8uP/nmv5UeXH/zzX8qA K/8AaWnf8/8A
a/8Af5f8aP7S07/n/tf+/wAv+NWPLj/55r+VHlx/881/KgCv/aWnf8/9r/3+ X/Gj+0tO/wCf
+1/7/L/jVjy4/wDnmv5UeXH/AM81/KgCv/aWnf8AP/a/9/l/xo/tLTv+f+1/ 7/L/AI1Y8uP/
AJ5r+VHlx/8APNfyoAr/ANpad/z/ANr/AN/l/wAapa7PZ3emyJb39mLmMiW3 JmUASIdy556E
jB9ia1fLj/55r+VHlx/881/Kk1dWZdOo6c1OO6Mr4Za5pU/xE8Q3xv7aKK40 HSWTzJVU587U
CV5PUZGR2r0r+2tG/wCgtYf+BKf415v8LbS2svix4ttmmhY3GlafPbwY+ZE8 +9Ln6b3J/wCB
AV6p5MP/ADyT/vkUoO61NMRTUKj5dnqvR7f5PzKf9taN/wBBaw/8CU/xo/tr Rv8AoLWH/gSn
+NXPJh/55J/3yKPJh/55J/3yKowKf9taN/0FrD/wJT/Gj+2tG/6C1h/4Ep/j VzyYf+eSf98i
jyYf+eSf98igCn/bWjf9Baw/8CU/xo/trRv+gtYf+BKf41c8mH/nkn/fIo8m H/nkn/fIoAp/
21o3/QWsP/AlP8aP7a0b/oLWH/gSn+NXPJh/55J/3yKPJh/55J/3yKAKf9ta N/0FrD/wJT/G
j+2tG/6C1h/4Ep/jVzyYf+eSf98ijyYf+eSf98igCn/bWjf9Baw/8CU/xo/t rRv+gtYf+BKf
41c8mH/nkn/fIo8mH/nkn/fIoArQ6tpc0qxQ6lZySMcKiTqST7AGrlVLyezs jAZVUNNMsMQV
MszH0+gBJ9ACe1W6LjcWkm+pgeO9Nk1PRfs9vLqUc7yJFG9leTW7R+YwRpCY 2XOxSzYPHy1v
1i+K/E2meG0tDqEqo13K0ce5gqKFQu7ux4VFVSSx6cdyK2gcjIoEefeKPDHx H8QeH9V0K58Z
+FIrTUrSa0kaPwvceYsciFCQTf43AHrjGe1NtvBXjCC3igXxZoRWNAgJ0GXJ wMf8/ddk+vaK
jFW1S0DA4IMo4NJ/b+if9BWz/wC/ooA5L/hD/GP/AENeg/8Aghl/+S6P+EP8 Y/8AQ16D/wCC
GX/5Lrrf7f0T/oK2f/f0Uf2/on/QVs/+/ooA5L/hD/GP/Q16D/4IZf8A5Lo/ 4Q/xj/0Neg/+
CGX/AOS663+39E/6Ctn/AN/RR/b+if8AQVs/+/ooA5L/AIQ/xj/0Neg/+CGX /wCS6P8AhD/G
P/Q16D/4IZf/AJLrrf7f0T/oK2f/AH9FH9v6J/0FbP8A7+igDkv+EP8AGP8A 0Neg/wDghl/+
S6P+EP8AGP8A0Neg/wDghl/+S663+39E/wCgrZ/9/RR/b+if9BWz/wC/ooA5 L/hD/GP/AENe
g/8Aghl/+S6P+EP8Y/8AQ16D/wCCGX/5Lrrf7f0T/oK2f/f0Uf2/on/QVs/+ /ooA5L/hD/GP
/Q16D/4IZf8A5Lo/4Q/xj/0Neg/+CGX/AOS663+39E/6Ctn/AN/RR/b+if8A QVs/+/ooA8y1
PwX418P67deOLPxb4fa4S0gt50fw/MVWCN5izcXgJAE7OV7mJMEc5l03xN8R pfEHiHR59Y8K
O2j3sdsJo9CuFEwe2hn3bTeHGPO29T93PfFejPruhOjI+p2TKwwymRSCPSvC tG8TeG/C3i/x
hYa34k0+Fm1OE2pkm5kt1sraOMk9ztQAn1BqNpev9f16HV/Eoecfyf8Ak/8A 0o9C/tjx/wD9
Bnwx/wCCSf8A+S6P7Y8f/wDQZ8Mf+CSf/wCS6zdJ8UeHdWsxeabrVld25Yp5 kUoI3A4I+oq3
/aum/wDP9b/99irOUn/tjx//ANBnwx/4JJ//AJLo/tjx/wD9Bnwx/wCCSf8A +S6g/tXTf+f6
3/77FH9q6b/z/W//AH2KAJ/7Y8f/APQZ8Mf+CSf/AOS6P7Y8f/8AQZ8Mf+CS f/5LqD+1dN/5
/rf/AL7FH9q6b/z/AFv/AN9igCf+2PH/AP0GfDH/AIJJ/wD5Lo/tjx//ANBn wx/4JJ//AJLq
D+1dN/5/rf8A77FH9q6b/wA/1v8A99igCf8Atjx//wBBnwx/4JJ//kuj+2PH /wD0GfDH/gkn
/wDkuoP7V03/AJ/rf/vsUf2rpv8Az/W//fYoAn/tjx//ANBnwx/4JJ//AJLo /tjx/wD9Bnwx
/wCCSf8A+S6g/tXTf+f63/77FVtV8R6LpmmXOoXWoW6wW8TSOfMHQDNJuyuy oxc5KMd2UNR8
U+MYPEWn3l5eaDeW1lqdjYukWmSxZe8uIoXIJuGw6Rygg4I/eMMdx7NXze/i 7wrfadoWlWXi
HT73VbnxFpM0kUEu5nkOpW8kmPYc49FX2r6QqYLS76m2JknPljtHRf5/N3fz MvxDoOna7CkV
+jnYsiBkbadkiGN1z6FWI/IjBANacaLHGsaKFVQAoHQAVzHxK0m61nQ47SyE 5uUl86BRFFJB
JIqttSdZAcxEnnbhhgEEHFdRVnOFFZz2epliV1hlBPA+zJxSfYtV/wCg03/g MlAGlRWb9i1X
/oNN/wCAyUfYtV/6DTf+AyUAaVFZv2LVf+g03/gMlH2LVf8AoNN/4DJQBpUV m/YtV/6DTf8A
gMlH2LVf+g03/gMlAGlRWb9i1X/oNN/4DJR9i1X/AKDTf+AyUAaVFZv2LVf+ g03/AIDJR9i1
X/oNN/4DJQBpUVm/YtV/6DTf+AyUfYtV/wCg03/gMlAGlWJq3/Ev8Q2OqjiG 6xYXR7ckmFj9
HLIPeb2qz9i1X/oNN/4DJVXVdFv9R06exn1pwkyFdy2yAqezA9iDgg9iBUzV 1ob4eooT97Z6
P0f+W680eZ+Bf+P7xd/2M9//AOhiunrmvD3w68TzjUL+x+Ic2nSXmoTz3lr/ AGRDKIrktiUK
zHJUsMr7EHvVb4c3et6x4Osr/UdWEl0zSxyOlsqhikrpnHbIUHFNO6ujOpTd ObhLdHXUVS+z
X3/QSP8A35Wj7Nff9BI/9+VpkF2iqX2a+/6CR/78rR9mvv8AoJH/AL8rQBdo ql9mvv8AoJH/
AL8rR9mvv+gkf+/K0AXaKpfZr7/oJH/vytH2a+/6CR/78rQBdrz/AOOLtc+B db0+M/JFps11
c/QKRGp+rZb/ALZmun1s6jZaLfXkepfPBbySrmFcZVSefyrnPDXwy8XeM/AE F7rnxBkgHiCx
jkvIIdKhOY3QYCvnI+U5wBwSepyTE9WonVh/cjKr20Xq/wDJXfrY99oqhb2m opMry6qZUB+Z
PIUbvxFX6s5TG8YeIIPDukPfSx+fIPuQhwpbHU5PQAck/QdSBWyORkVneINF 0bXLL7JrenWt
7AeAs6A4J9D1BPHStBVCqFUAKBgADgUAZ761ZKxUpf5BwcWE5H57KT+3LH/n nf8A/gvn/wDi
K0qKAM3+3LH/AJ53/wD4L5//AIij+3LH/nnf/wDgvn/+IrSooAzf7csf+ed/ /wCC+f8A+Io/
tyx/553/AP4L5/8A4itKigDN/tyx/wCed/8A+C+f/wCIo/tyx/553/8A4L5/ /iK0qKAM3+3L
H/nnf/8Agvn/APiKP7csf+ed/wD+C+f/AOIrSooAzf7csf8Annf/APgvn/8A iKP7csf+ed//
AOC+f/4itKigDN/tyx/553//AIL5/wD4ij+3LH/nnf8A/gvn/wDiK0qKAM3+ 3LH/AJ53/wD4
L5//AIij+3LH/nnf/wDgvn/+IrSooA5MazZad4pZwl8LXVVG7NjMMXMa8fwZ JeMfh5I9a8y+
Et/bxeA7NGW5yJ7o/LbSMObiQ9QuK9q8QWDalpUttFIIrgYkt5SM+XKhDI3u AwGR3GR3rx34
NsX+HlgzKVYzXRKnsftMvFRHSTXzOqr+8pRqdVo/0/DT5eZ0f9p239y7/wDA SX/4mj+07b+5
d/8AgJL/APE1doqzlKX9p239y7/8BJf/AImj+07b+5d/+Akv/wATV2igCl/a dt/cu/8AwEl/
+Jo/tO2/uXf/AICS/wDxNXaKAKX9p239y7/8BJf/AImj+07b+5d/+Akv/wAT V2koA5Hx/qkc
+gXOm2/2lZLm3laQm2kG2FVJfqvQnantvz2ru/hfrFnF8M/C0bJfFk0a0U7b GZhkQp0ITB+o
ri7v/TPDuuay3IurOVbb2gVG2n/gRLP9GUdq9E+FP/JLvCf/AGBbP/0QlRDX 3u51Yn3LUf5d
/V7/AHaL5X6mtb6vZzzLEiXoZjgb7KZR+JKgD8av0UVZynN/EXR5Ne8OS6Zb 2okuZcmCZo4n
SCQDKs4fnaSNpKfOATtIPI6SsfxfqGqaZo8l5pNnb3k8fIglkKGY9o0wD8zH CgngE81sUAZz
6deMxI17UlBOQAlvge3MVJ/Zt7/0MGp/9+7f/wCNUr6vErFfsWpHBxkWchH8 qT+2Yf8Any1P
/wAA5P8ACgA/s29/6GDU/wDv3b//ABqj+zb3/oYNT/792/8A8ao/tmH/AJ8t T/8AAOT/AAo/
tmH/AJ8tT/8AAOT/AAoAP7Nvf+hg1P8A792//wAao/s29/6GDU/+/dv/APGq P7Zh/wCfLU//
AADk/wAKP7Zh/wCfLU//AADk/wAKAD+zb3/oYNT/AO/dv/8AGqP7Nvf+hg1P /v3b/wDxqj+2
Yf8Any1P/wAA5P8ACj+2Yf8Any1P/wAA5P8ACgA/s29/6GDU/wDv3b//ABqj +zb3/oYNT/79
2/8A8ao/tmH/AJ8tT/8AAOT/AAo/tmH/AJ8tT/8AAOT/AAoAP7Nvf+hg1P8A 792//wAao/s2
9/6GDU/+/dv/APGqP7Zh/wCfLU//AADk/wAKP7Zh/wCfLU//AADk/wAKAD+z b3/oYNT/AO/d
v/8AGqP7Nvf+hg1P/v3b/wDxqj+2Yf8Any1P/wAA5P8ACj+2Yf8Any1P/wAA 5P8ACgA/s29/
6GDU/wDv3b//ABqj+zb3/oYNT/792/8A8ao/tmH/AJ8tT/8AAOT/AAo/tmH/ AJ8tT/8AAOT/
AAoAP7Nvf+hg1P8A792//wAarzGL4aDwvdrYad428T2mm3zTSRYWyfZc8vsJ e2b5WUHHTBQ5
JLCvTv7Zh/58tT/8A5P8Kz/EVxb6rotzYtZ6ojOmYnWyfdHIvzI446hgD+FT NXV1ujfD1Ixk
4z+F6P8Az+T189jznwFJqWr+BtA1a91i7e6vdMtriZljhALvErMQAnHJNbX2 O5/6C17/AN8x
f/EVwHg/xlF4e+Hun22q+FvF9qNE0yGHUHfRZitv5UCly5AwoCjdz2INd/8A 2jH/AM+17/4D
v/hTTTV0Z1KcqcnGW6D7Hc/9Ba9/75i/+Io+x3P/AEFr3/vmL/4ij+0Y/wDn 2vf/AAHf/Cj+
0Y/+fa9/8B3/AMKZAfY7n/oLXv8A3zF/8RR9juf+gte/98xf/EUf2jH/AM+1 7/4Dv/hR/aMf
/Pte/wDgO/8AhQAfY7n/AKC17/3zF/8AEVl67bXU5h0iPVbwvebhL8sQ2QDH mHhMgnIUehYH
sa0bjVra3gknniu44o1Lu7W7AKAMkniuR0HxlFdT3+ox+GfFt7I9w9uz2uiT yxxiJivlhgME
g7i3+0WHQConr7p1YdezvWfTb16fdv8Ag9zX+G3g/UfGPwz0vVdU8eeJIn1O 0JmhtoNPSNQS
V2qDakgY46k+9elaL4cfSNGstJstf1VbWyt47eEMtuSERQq5PlcnAFcz8E57 jSPhT4d07U9I
1e1vILMCWGSxkDISScEY4PNdl/bMP/Plqf8A4Byf4VZyj7ewuo5lkfWr+ZVO TG6QBW9jtjB/
I1fqhb6pFNMsQtL9Cxxue1dVH1JHFX6AMzxDpmj6nbwx6zFFJEkwMQkkKDzG +UYwRkncQB74
71pgBQAAABwAKxPGul3Wr6Rb2tmIfMj1Oxuj5rFRsguopnxgHkrGQPcjkda2 6ACism2sbS+t
or211W+mt7hBLFJFeMUdGGQVI4IIPFSf2On/AEENT/8AApqANKis3+x0/wCg hqf/AIFNR/Y6
f9BDU/8AwKagDSorN/sdP+ghqf8A4FNR/Y6f9BDU/wDwKagDSorN/sdP+ghq f/gU1H9jp/0E
NT/8CmoA0qKzf7HT/oIan/4FNR/Y6f8AQQ1P/wACmoA0qKzf7HT/AKCGp/8A gU1H9jp/0ENT
/wDApqANKis3+x0/6CGp/wDgU1H9jp/0ENT/APApqANKis3+x0/6CGp/+BTU f2On/QQ1P/wK
agDSorN/sdP+ghqf/gU1H9jp/wBBDU//AAKagDjPjzZXUPw28Xalp0RkNx4f vLa9hBA3oYJA
kgzxuQnn1Ut1IUU5HWRFdGDKwyrA5BHqK6TXvClhreh3+jahe6o9nf20lrcK LtgWjkUqwz9C
a8f+I/gJ/A+kwappfiTxS1jHe2Ea+Zq7eXbBruJJBIpwGjaNmAxyD6g/Lm/c d+h2RX1iKj9t
bea7eq6d1p0SffUVS/s9P+fu8/7/AJo/s9f+fu9/7/mtDjLtFUv7PX/n7vf+ /wCa5L/hG7vx
T8SdT0e21/xBbQWmm2Dl7TU2jjt2eW680uoOWkZUiCjj1PAwZlJRRrSpSqys vm+iXd/15LU6
K5P9r6l9jXmxtHDXDdpZRysfuF4ZvfaP7wra+BH/ACKGpf8AYw6r/wClstU4 fg1osEflw+Kv
GsaZJ2prTqMk5JwB3JJro/CfgfTPDOjjS9N1DWTD50s7PNfM7vJI5d2Y9yWY miMbavcqvUUr
Rh8K2/zfm/8AJdDqKKzf7HT/AKCGp/8AgU1H9jp/0ENT/wDApqowNKistrS0 08x3Nxqd2iCV
Ix590djO7BEXnqSzAAdyQK1KAOX8f+I7zw/bxtZQQO7W13cs04OzEEJk2cEY LHHPYBuDXSW0
vnW8U2xo/MQNtYcrkZwfemXlnaXsax3lrBcorblWWMOAcEZAPfBI/E1PQB5x ol/4v0n4Z2Oi
WvgbXU1iz0aO1ik+0aeYxcJCFB5ueV3DuOnatTWNf8WNo+3TPBOvJf7ovmeb T9uN67+tweq7
q7OigDj9Z1/xS0dt/ZngjXkcXURn3zafzDuHmAZuDyVzRqOv+KW1DTGsvBGv LarO5vVabT8t
H5ThQP8ASOu/YeMcA12FFAHHz6/4pOv2jxeCNeGmC1nFwhm0/cZi8PlEf6Rn AUTZ57jrxgg1
/wAUjX7t5fBGvHTDawC3QTafuEwebzSf9IzgqYcc9j05z2FFAHH2Gv8Aikan qLXfgjXms2dP
sarNp+VXYN2f9I/vZ65o0fX/ABSq3f8AafgjXnJupDb7JtP4hz8gOLgc4rsK KAOO0XX/ABWm
mMuqeCdeku/OnIZJtPx5ZlcxDi4HITYD7jvVN9d8df8ACBmFfBmt/wDCS/2X tE3naf5f2zys
bv8Aj4xt8znpjHau9ooA4/Xdf8UvYKNJ8Ea9Hc/aYCxebT8eUJkMo5uDyY94 HuR060avr/il
msv7O8Ea8gF0hut02n/NDg7gM3HXO2uwooA4++1/xSdV09rTwRry2StJ9sUz aflhs+TH+kf3
vpSTa/4rPiK1eLwTrw0oWkwuEM2n7jOXi8oj/SM4CiXPPcde3Y0UAcfBr/ik a/dvL4I146Yb
WAW6CbT9wmDzeaT/AKRnBUw457Hpzk03X/FK3+pm+8Ea81s1ypsQs2n5WLyY wwP+kdfMEh5z
wR9K7CigDj9H1/xSq3f9p+CNecm6kNvsm0/iHPyA4uBzijQdf8UppUS6v4I1 6S9DPvZJtPwR
vO3pcD+HFdhRQBwH9u+PP+EA8n/hDNb/AOEm/srb53naf5f2zysbsfaMbfM5 6Yx2qj8WZfE/
iLwTLpel+Atde6N9Y3AWS5sFDJDeQzOMm5xnZG2PfHSvTaKAPBrmP4i2Mtr/ AMI/8N/Ei2iz
Zmsrm90sJs2t8sbi6YoNxU7cEDHGBwdqR/Hcl1Yyp4B1uCBd5u4jd6exbK/K FP2js30r1+io
5LbOx1PFc2tSKk+7vf52av6vXzPDNQk+JtxrXlp8PPENrpQiZWNve6Y80rZX BG65GzjcO/X1
wRq/DuDxJoni/W9Tk+G/iCzsbvTbG2hQ3envI0kUt28jNi6Oc+enJJJO7Pqf XqKaik79SKle
U48qVo9l/V3827dDj9M1/wAUre6ob7wRrz27XSmwCzaflIfJiBB/0jr5olPO eCPoDRNf8Upb
TjVPBGvSSm6nMRSbT8CEyN5Q4uByE2g+/rXYUVRgcXpGveLl8PQpqXgnXn1Q QkSOs2n7TJzg
8XGMdO1QT6743PgsxQ+C9cHiD7AFEpm0/wAv7Ts5b/j4xjd7Y9q7uigDiPEd 7rut2lnp8Xgr
XLX/AImlhO81xPZeXHHFdxSux2XDMcKjcAEmu3oooA//2Q==
--------------090600040301020502030508--
|
|
|
|
| Re: cubase mixing levels [message #75023 is a reply to message #75017] |
Sat, 28 October 2006 10:55 |
chuck duffy
Messages: 453
Registered: July 2005
|
Senior Member |
|
|
See now it get's even more confusing....
Why? Check any math text, sine wavs are always represented with signed numbers,
doesn't matter if it is integer or float.
Even if there is no sign bit on the platform, we fake it by splitting the
highest possible value for a word size in half just to link what we are doing
to the math world.
Positive amplitudes are positive numbers, negative amplitudes are negative
numbers, and zero, most definitely is silence :-)
So... In a sense you are right, but your graph is wrong :-)
Chuck
"Dave(EK Sound)" <audioguy_editout_@shaw.ca> wrote:
>
>Ah, now, this is where people get confused.
>
>The assumption that FULL bits (all 1111's) represents 0dbfs,
>and that empty bits (all 00000's) represents the noise floor
>is false.
>
>The full bits actually represent the maximum positive
>amplitude, and the empty bits represent the maximum
>*negative* amplitude of the bi-phase audio signal. (I have
>attached a pic below of a 4 bit signal capture, the vertical
>axis shows the 4 bits, while the horizontal axis shows the
>sample rate)
>
>So, what is actually happening in 24 bit vs 16 bit is that
>there are more bits to represent the vertical axis. This
>means that "no signal" on the input of the recorder would
>yield a number half way up, not all zeros.
>
>Does this help?
>
>David.
>
>Dedric Terry wrote:
>
>> Hi Ted,
>>
>> 16 bit and 24 bit both only represent up to 0dB full scale (FS). The
>> dynamic range afforded by 24-bit extends down to -144dB rather than the
>> -96dB of 16 bit. That's what we are interested in with digital audio,
not
>> the theoretical limits above 0.
>>
>> Once in the DAW, -48dB is represented by the lower 8 bits in 16-bit words,
>> and the lower 16 bits in 24-bit words. That's probably how they came
to
>> that conclusion, but it's mathematically incorrect since the same bit
>> actually represents -48dB, you just add an extra 8 onto the bottom of
the
>> dynamic range, not the top as the quote seems to assume.
>>
>> Regards,
>> Dedric
>>
>> On 10/28/06 3:51 AM, in article 4543283c$1@linux, "Ted Gerber"
>> <tedgerber@rogers.com> wrote:
>>
>>
>>>Your quote here:
>>>
>>>" if you happen to record your tracks at lower levels (e.g.
>>>
>>>>-10dB peaks, then your peaks are now at -32dB, which is 1/3 the resolution
>>>>of 16 bit audio - that isn't insignificant). "
>>>
>>>The other statement said that 24bit recording at -48db is equal to a full
>>>range
>>>16bit recording...
>>>
>>>These 2 statements look like they're talking about the same thing (apples
>>>to apples). If they are, how are they reconciled? If they're not and I'm
>>>misunderstanding...
>>>
>>>Ted
>>
>>
>
|
|
|
|
| Re: cubase mixing levels [message #75026 is a reply to message #75023] |
Sat, 28 October 2006 10:16 |
audioguy_editout_
Messages: 249
Registered: December 2005
|
Senior Member |
|
|
Yah, I was refering to the sampling phase of straight PCM
encoding. I don't even want to know what goes on after
that! Geek!! ;-)
David.
chuck duffy wrote:
> See now it get's even more confusing....
>
> Why? Check any math text, sine wavs are always represented with signed numbers,
> doesn't matter if it is integer or float.
>
> Even if there is no sign bit on the platform, we fake it by splitting the
> highest possible value for a word size in half just to link what we are doing
> to the math world.
>
> Positive amplitudes are positive numbers, negative amplitudes are negative
> numbers, and zero, most definitely is silence :-)
>
> So... In a sense you are right, but your graph is wrong :-)
>
> Chuck
>
>
>
>
>
> "Dave(EK Sound)" <audioguy_editout_@shaw.ca> wrote:
>
>>Ah, now, this is where people get confused.
>>
>>The assumption that FULL bits (all 1111's) represents 0dbfs,
>>and that empty bits (all 00000's) represents the noise floor
>>is false.
>>
>>The full bits actually represent the maximum positive
>>amplitude, and the empty bits represent the maximum
>>*negative* amplitude of the bi-phase audio signal. (I have
>>attached a pic below of a 4 bit signal capture, the vertical
>>axis shows the 4 bits, while the horizontal axis shows the
>>sample rate)
>>
>>So, what is actually happening in 24 bit vs 16 bit is that
>>there are more bits to represent the vertical axis. This
>>means that "no signal" on the input of the recorder would
>>yield a number half way up, not all zeros.
>>
>>Does this help?
>>
>>David.
>>
>>Dedric Terry wrote:
>>
>>
>>>Hi Ted,
>>>
>>>16 bit and 24 bit both only represent up to 0dB full scale (FS). The
>>>dynamic range afforded by 24-bit extends down to -144dB rather than the
>>>-96dB of 16 bit. That's what we are interested in with digital audio,
>
> not
>
>>>the theoretical limits above 0.
>>>
>>>Once in the DAW, -48dB is represented by the lower 8 bits in 16-bit words,
>>>and the lower 16 bits in 24-bit words. That's probably how they came
>
> to
>
>>>that conclusion, but it's mathematically incorrect since the same bit
>>>actually represents -48dB, you just add an extra 8 onto the bottom of
>
> the
>
>>>dynamic range, not the top as the quote seems to assume.
>>>
>>>Regards,
>>>Dedric
>>>
>>>On 10/28/06 3:51 AM, in article 4543283c$1@linux, "Ted Gerber"
>>><tedgerber@rogers.com> wrote:
>>>
>>>
>>>
>>>>Your quote here:
>>>>
>>>>" if you happen to record your tracks at lower levels (e.g.
>>>>
>>>>
>>>>>-10dB peaks, then your peaks are now at -32dB, which is 1/3 the resolution
>>>>>of 16 bit audio - that isn't insignificant). "
>>>>
>>>>The other statement said that 24bit recording at -48db is equal to a full
>>>>range
>>>>16bit recording...
>>>>
>>>>These 2 statements look like they're talking about the same thing (apples
>>>>to apples). If they are, how are they reconciled? If they're not and I'm
>>>>misunderstanding...
>>>>
>>>>Ted
>>>
>>>
>
|
|
|
|
| Re: cubase mixing levels [message #75033 is a reply to message #75023] |
Sat, 28 October 2006 12:09 |
Dedric Terry
Messages: 788
Registered: June 2007
|
Senior Member |
|
|
I think I have clouded the issue. No, wait, I know I have. ;-)
Chuck is right about how bits are translated to amplitude representations,
and I knew this to be the case, but when talking about dynamic range we are
talking about power in terms of dBFS. Here we don't refer to the +/- aspect
of amplitude, but rather signal power only relative to 0dBFS (power is an
absolute value since negative power is only theoretical); but, since the
maximum value chosen for digital is 0dBFS, we use negative dB values to
represent the power range of the signal.
So in actuality, all 1's (sign bit excluded) represent the maximum amplitude
of the signal which can be a + or - amplitude, in absolute value, (which
translates to 0dB Full Scale digital, but not 0db!), and all 0's represent
0db actual power level ("silence", aka anything below -96dB for 16 bit
and -144dB for 24-bit - correct Chuck?).
Since audio is quantized into the word based on voltage, not power, 1.1 volt
(I believe this is standard) is given the reference level of 0dBFS at the
ADC. So in effect, 24-bits provides better resolution as we approach 0
volts, and 0db in signal power, than 16 bits. The dynamic range is a
relative value. 16 or 24-bit words are actual signed values for voltages
between 0 and 1.1 volts. The area under that point in the curve is where we
get power but that is inverted to create the dB digital range in steps
of -6dB per bit from 0dBFS down. So, adding 8 bits to a 16 bit word doesn't
decrease the power step size to less than 6dB per bit, but rather extend the
power range down an extra -48dB to -144dB - just a matter of detecting lower
voltage levels at the ADC so we get better sensitivity in the recording, not
the headroom to record louder drummers. ;-)
Do you guys agree with this explanation or is that even more convoluted?
Regards,
Dedric
"chuck duffy" <c@c.com> wrote in message news:45438b69$1@linux...
>
> See now it get's even more confusing....
>
> Why? Check any math text, sine wavs are always represented with signed
> numbers,
> doesn't matter if it is integer or float.
>
> Even if there is no sign bit on the platform, we fake it by splitting the
> highest possible value for a word size in half just to link what we are
> doing
> to the math world.
>
> Positive amplitudes are positive numbers, negative amplitudes are negative
> numbers, and zero, most definitely is silence :-)
>
> So... In a sense you are right, but your graph is wrong :-)
>
> Chuck
>
>
>
>
>
> "Dave(EK Sound)" <audioguy_editout_@shaw.ca> wrote:
>>
>>Ah, now, this is where people get confused.
>>
>>The assumption that FULL bits (all 1111's) represents 0dbfs,
>>and that empty bits (all 00000's) represents the noise floor
>>is false.
>>
>>The full bits actually represent the maximum positive
>>amplitude, and the empty bits represent the maximum
>>*negative* amplitude of the bi-phase audio signal. (I have
>>attached a pic below of a 4 bit signal capture, the vertical
>>axis shows the 4 bits, while the horizontal axis shows the
>>sample rate)
>>
>>So, what is actually happening in 24 bit vs 16 bit is that
>>there are more bits to represent the vertical axis. This
>>means that "no signal" on the input of the recorder would
>>yield a number half way up, not all zeros.
>>
>>Does this help?
>>
>>David.
>>
>>Dedric Terry wrote:
>>
>>> Hi Ted,
>>>
>>> 16 bit and 24 bit both only represent up to 0dB full scale (FS). The
>>> dynamic range afforded by 24-bit extends down to -144dB rather than the
>>> -96dB of 16 bit. That's what we are interested in with digital audio,
> not
>>> the theoretical limits above 0.
>>>
>>> Once in the DAW, -48dB is represented by the lower 8 bits in 16-bit
>>> words,
>>> and the lower 16 bits in 24-bit words. That's probably how they came
> to
>>> that conclusion, but it's mathematically incorrect since the same bit
>>> actually represents -48dB, you just add an extra 8 onto the bottom of
> the
>>> dynamic range, not the top as the quote seems to assume.
>>>
>>> Regards,
>>> Dedric
>>>
>>> On 10/28/06 3:51 AM, in article 4543283c$1@linux, "Ted Gerber"
>>> <tedgerber@rogers.com> wrote:
>>>
>>>
>>>>Your quote here:
>>>>
>>>>" if you happen to record your tracks at lower levels (e.g.
>>>>
>>>>>-10dB peaks, then your peaks are now at -32dB, which is 1/3 the
>>>>>resolution
>>>>>of 16 bit audio - that isn't insignificant). "
>>>>
>>>>The other statement said that 24bit recording at -48db is equal to a
>>>>full
>>>>range
>>>>16bit recording...
>>>>
>>>>These 2 statements look like they're talking about the same thing
>>>>(apples
>>>>to apples). If they are, how are they reconciled? If they're not and I'm
>>>>misunderstanding...
>>>>
>>>>Ted
>>>
>>>
>>
>
|
|
|
|
|
|
|
|
| Re: cubase mixing levels [message #75047 is a reply to message #75041] |
Sat, 28 October 2006 14:54 |
Dedric Terry
Messages: 788
Registered: June 2007
|
Senior Member |
|
|
Yes with amplitude you get greater resolution with higher bit depths (more
levels to quantize each sample). This isn't the same as the dynamic range
though.
For each extra bit added, we basically reduce the quantization noise by
6dB - we halve the distance between quantization levels (doubling the number
of quantization levels for that same range of amplitude - 0v to 1.1v).
Since power ratio (dB) is a log function we get a consistent change of
relative power of 6dB for each halving of the distance between quantization
levels (when referring to the quantization noise floor - 6dB is also often
used to refer to "twice the loudness" - but whether you are referring to 6dB
as lowering noise or raising volume the power change is the same, but with
digital audio we are referring to lowering quantization noise, not doubling
the "loudness" with each bit).
So, dB is a measurement of power ratios (change in power).
dB = 10*log10(P2/P1)
where P1 is the power being measured, and P1 is the reference to which P2 is
being compared.
Likewise, voltage ratios are:
A = 20*log10(V2/V1)
So the dynamic range increases by 6dB with every bit we add (by lowering
quantization noise with twice the quantization levels), and in the digital
realm, the reference is 0dBFS so the range (really "depth") extends down
from there as we push the quantization noise lower and lower (hence -144dB
for 24 bit). Amplitude level changes are more accurately represented by
more bits since we have more levels to depict each level between 0v and
1.1v, but the 0 to 1.1volt range never changes in the converter whether
it's a 4 bit or 24 bit converter (of course the max 1.1 volts may vary
depending on the converter application, design and analog input transform,
but with audio I believe 1.1v is most common).
Anyone else feel free to add or expand on this. I may have still missed
some informative bit of techdom, or mis-stated something, considering how I
originally turned this murky water into thick mud. ;-)
Dedric
"Neil" <IOUOIU@OIU.com> wrote in message news:4543b952@linux...
>
> "Dedric Terry" <dedric@echomg.com> wrote:
>>Do you guys agree with this explanation or is that even more convoluted?
>
> I'm totally confused now - I thought it was linear...
> everything else in the digital world is until you get into
> floating-point stuff, innit? so why not the number of bits in a
> given equal-amplitude (vertical) segment of the signal vs the
> next segment of the same amplitude differential?
>
> Neil
|
|
|
|
|
|
| Re: cubase mixing levels [message #75049 is a reply to message #75047] |
Sat, 28 October 2006 15:21 |
AlexPlasko
Messages: 211
Registered: September 2006
|
Senior Member |
|
|
dedric:
so if we use 32 bit floating point math, the quantization error created by
the software data processing will be eliminated?or were you referring to
errors created at the converters?
"Dedric Terry" <dedric@echomg.com> wrote in message news:4543cf8b$1@linux...
> Yes with amplitude you get greater resolution with higher bit depths (more
> levels to quantize each sample). This isn't the same as the dynamic range
> though.
>
> For each extra bit added, we basically reduce the quantization noise by
> 6dB - we halve the distance between quantization levels (doubling the
> number of quantization levels for that same range of amplitude - 0v to
> 1.1v). Since power ratio (dB) is a log function we get a consistent change
> of relative power of 6dB for each halving of the distance between
> quantization levels (when referring to the quantization noise floor - 6dB
> is also often used to refer to "twice the loudness" - but whether you are
> referring to 6dB as lowering noise or raising volume the power change is
> the same, but with digital audio we are referring to lowering quantization
> noise, not doubling the "loudness" with each bit).
>
> So, dB is a measurement of power ratios (change in power).
>
> dB = 10*log10(P2/P1)
> where P1 is the power being measured, and P1 is the reference to which P2
> is being compared.
>
> Likewise, voltage ratios are:
> A = 20*log10(V2/V1)
>
> So the dynamic range increases by 6dB with every bit we add (by lowering
> quantization noise with twice the quantization levels), and in the digital
> realm, the reference is 0dBFS so the range (really "depth") extends down
> from there as we push the quantization noise lower and lower (hence -144dB
> for 24 bit). Amplitude level changes are more accurately represented by
> more bits since we have more levels to depict each level between 0v and
> 1.1v, but the 0 to 1.1volt range never changes in the converter whether
> it's a 4 bit or 24 bit converter (of course the max 1.1 volts may vary
> depending on the converter application, design and analog input transform,
> but with audio I believe 1.1v is most common).
>
> Anyone else feel free to add or expand on this. I may have still missed
> some informative bit of techdom, or mis-stated something, considering how
> I originally turned this murky water into thick mud. ;-)
>
> Dedric
>
> "Neil" <IOUOIU@OIU.com> wrote in message news:4543b952@linux...
>>
>> "Dedric Terry" <dedric@echomg.com> wrote:
>>>Do you guys agree with this explanation or is that even more convoluted?
>>
>> I'm totally confused now - I thought it was linear...
>> everything else in the digital world is until you get into
>> floating-point stuff, innit? so why not the number of bits in a
>> given equal-amplitude (vertical) segment of the signal vs the
>> next segment of the same amplitude differential?
>>
>> Neil
>
>
|
|
|
|
| Re: cubase mixing levels [message #75050 is a reply to message #75011] |
Sat, 28 October 2006 16:16 |
Ted Gerber
Messages: 705
Registered: January 2009
|
Senior Member |
|
|
Thanks Dedric-
Either the author made the mistake you've postulated, or the author
was misqouted.
Thanks again, the people on this newsgroup (like you) make it the great
place to hang out that it is - virtually speaking : )
Ted
Dedric Terry <dterry@keyofd.net> wrote:
>Hi Ted,
>
>16 bit and 24 bit both only represent up to 0dB full scale (FS). The
>dynamic range afforded by 24-bit extends down to -144dB rather than the
>-96dB of 16 bit. That's what we are interested in with digital audio, not
>the theoretical limits above 0.
>
>Once in the DAW, -48dB is represented by the lower 8 bits in 16-bit words,
>and the lower 16 bits in 24-bit words. That's probably how they came to
>that conclusion, but it's mathematically incorrect since the same bit
>actually represents -48dB, you just add an extra 8 onto the bottom of the
>dynamic range, not the top as the quote seems to assume.
>
>Regards,
>Dedric
>
>On 10/28/06 3:51 AM, in article 4543283c$1@linux, "Ted Gerber"
><tedgerber@rogers.com> wrote:
>
>> Your quote here:
>>
>> " if you happen to record your tracks at lower levels (e.g.
>>> -10dB peaks, then your peaks are now at -32dB, which is 1/3 the resolution
>>> of 16 bit audio - that isn't insignificant). "
>>
>> The other statement said that 24bit recording at -48db is equal to a full
>> range
>> 16bit recording...
>>
>> These 2 statements look like they're talking about the same thing (apples
>> to apples). If they are, how are they reconciled? If they're not and I'm
>> misunderstanding...
>>
>> Ted
>
|
|
|
|
| Re: cubase mixing levels [message #75053 is a reply to message #75047] |
Sat, 28 October 2006 17:01 |
Neil
Messages: 1645
Registered: April 2006
|
Senior Member |
|
|
"Dedric Terry" <dedric@echomg.com> wrote:
>Yes with amplitude you get greater resolution with higher bit
>depths (more levels to quantize each sample).
Yes, that's what I meant, and it's linear, right? In other
words, if a sine wave is full-on 0db flat-out at peaks, then
(in a simplistic example) 12 bits are taken up in half that
amplitude, and 12 bits are taken up in the remaining half, or 6
bits are taken up in a quarter of it, and so on, right?
>This isn't the same as the dynamic range though.
Ooooooooh, ok, now I'm confused again.
If you're recording at a low level... let's say peaks of -40
or -50 db on a given track, chances are you're only using 12 or
so bits of that file capacity even though you may
be "recording" at 24-bits, yes? Or are you saying becasue db is
log, not linear, you might actually be only using 6 or so of
your available bits if you're recording at those low peaks of
-40 or -50?
Neil
|
|
|
|
| Re: cubase mixing levels [message #75065 is a reply to message #74995] |
Sat, 28 October 2006 18:37 |
Aaron Allen
Messages: 1988
Registered: May 2008
|
Senior Member |
|
|
If all mix busses are the same, why did the PT III rigs suck out loud so
badly unless you kicked the master fader down to -15db/-20db?
AA
"Dedric Terry" <dterry@keyofd.net> wrote in message
news:C1684104.4BF6%dterry@keyofd.net...
>I agree with Martin completely. To add my own general opinion on this long
> running "sound of summing" debate, that is fast become urban legend:
>
> I think it is time to start debunking some of the fables around digital
> summing. Paris cutting levels and then adding the gain back at the master
> does one and only one thing: pushes all tracks down by 22dB to make it
> easier to sum them well below 0dBFS, and then make some of it back before
> the master so you didn't know what happened. If you are mixing properly
> in
> any native DAW, you will in effect do the same thing - lower track levels
> such that the peak of the summation remains below 0dBFS.
>
> In simple math terms:
>
> 1- Native DAW: 2+2=4
> 2- Paris: ((2-1)+(2-1)) + 2 = 4
>
> With SX you adjust the gain yourself on the way in, if you like, or better
> yet, set the levels for the mix at hand as needed, as you go.
>
> So why do this in Paris? I am guessing Paris had to convert all audio to
> 24-bit if sent to the native cpu for native plugins in order to prevent
> clipping before you even start dropping faders on the mix (unless they
> somhow converted to 32-bit float on the EDS chip first, which isn't a
> 32-bit
> float chip, so that seems unlikely). (No one would want to mix with most
> faders at -40dB - it would "seem" wrong). 24-bits for any portion of
> summing (not tracking) is a limitation esp. if tracked audio files are
> near
> 0dBFS to begin with - there is no where to go with gain addition, and only
> subtraction to work with. You can add a lot more -22dB peak audio files
> to
> a mix without clipping (with all faders at 0), where only 2 audio files
> peaking just below 0dBFS will automatically clip. At least that seems to
> be
> the reasoning behind it, but imho, only applicable to prevent a potential
> problem in the DAW itself, not as prescribed digital audio practice.
>
> Back to native DAWs: While this approach may seem somehow capable of
> producing a different sound to a final mix, the only think you gain by
> lowering tracks by 22dB from the start is lower bit resolution. This is
> especially true if you happen to record your tracks at lower levels (e.g.
> -10dB peaks, then your peaks are now at -32dB, which is 1/3 the resolution
> of 16 bit audio - that isn't insignificant). So, if you combine the
> flawed
> advice to record at -10 to -20dB, with the concept of lowering all tracks
> by
> 22dB before you start mixing, and you end up removing most of the
> resolution
> we work hard for with quality mics, preamps and converters, and mixing as
> much more noise and quantization error than necessary.
>
> As Martin suggested, record just below, or comfortably below clipping (as
> the source dictates) to maximize your use of bit resolution - e.g. keep
> audible that ambience or depth of the recording that tails off down at
> those
> lower levels, rather than mixing it with mic self-noise, etc.
>
> John, what I think the reference you quoted is actually saying (or should
> be
> at least) is just that 24-bit digital (with a quality front end and
> converters) affords a higher signal to noise ratio than analog did, so
> pushing record levels to widen the gap between peaks and the noise floor
> isn't as critical. But since quite a few really great mics have a noise
> floor of around -70 to -74dB, we are still putting noise into half of the
> bits of that glorious 24-bit range.
>
> In terms of signal vs. noise - why record less signal when you can record
> more?
>
> Imho, there are a lot of "famous" engineers out there spouting complete
> technical rubbish out of lack of true knowledge, or passing along
> conversational heresay and conjecture. Sadly, engineering is becoming
> more
> about letting the gear dictate the recording process than the engineer,
> and
> I believe that's what's wrong with music today. Too my people think xyz
> piece of gear can make a hit, a vibe, or a certain sound just because
> someone else did it, but few actually use their skills to create the sound
> by knowing what combinations of gear will help get them there.
>
> 9/10 times it isn't about obtaining one piece of gear to get "that sound",
> but understanding the 1000 different possibilities and knowing how to use
> any of them.
>
> The Nuendo and SX audio engines are identical. They also are identical in
> summing to Sequoia/Samplitude, and probably Logic, Audition, and DP too
> (I've tested Nuendo and Sequoia beyond the limits of normal recording to
> verify this, so this comes from experience and listening, not speculation
> or
> internet urban legend).
>
> Regards,
> Dedric
>
> On 10/27/06 10:03 PM, in article 4542d488@linux, "alex plasko"
> <alex.plasko@snet.net> wrote:
>
>> hi martin
>> I think what john is referring to is what started this thread. We are
>> trying to emulate the way paris handles files at mixdown, not at
>> recording
>> files, at mixdown.
>> chuck said that paris automatically and transparently cuts channel levels
>> by -22db, and then adds it back automatically when it hits the submix bus
>> much the way analog consoles do.
>> what we were toying with is if was possible to emulate that *effect*
>> with
>> other daws by cutting channel levels 22db and making it back up at the
>> output bus.
>> what we dont know is how cubase/nuendo mix bus handles the files.or
>> exactly
>> how paris does it for that matter.
>> If we can duplicate the way paris handles the mix bus DJ can sleep nights
>> again .
>> "Martin Harrington" <lendan@bigpond.net.au> wrote in message
>> news:4542c966$1@linux...
>>> Can't tell you anything technically about the Cubase mix bus, (I use
>>> Nuendo),but I think it's basically the same, but it's a fallacy that if
>>> you record at lower levels you are protecting the file from clipping.
>>> What you are doing is not using all the "bits" available to you, and
>>> therefore start introducing unwanted artifacts into the mix.
>>> If the "bits" aren't there on the original recording, and the levelis
>>> cosewuently low at the mix bus, no matter what you do, you can't get
>>> those
>>> bits back and the resolution and "size" of your mix has to suffer.
>>> I record as hot as I can, and use the channel faders to mix, usually
>>> never
>>> moving the master fader, although having said that, my mixes for TV/
>>> Doco
>>> work are not quite as complicated as most decent size music mixes would
>>> be.
>>> --
>>> Martin Harrington
>>> www.lendanear-sound.com
>>> "John" <no@no.com> wrote in message news:45429eda@linux...
>>>> Martin, so do you know anything about the Cubase mix bus? Do they
>>>> maybe
>>>> mean that on mixdown you pull the faders way back but still record hot?
>>>> Just wondering how the Cubase mix bus behaves.
>>>>
>>>> Thanks
>>>>
>>>> Martin Harrington wrote:
>>>>> That's not true, and dont let anyone tell you it is.
>>>>> You still need to get all levels as optimised as possible, as we all
>>>>> did
>>>>> with tape.
>>>>> Otherwise you are not using all the bits available to you, and noise
>>>>> will be the end result.
>>>>> This is why the good/great engineers are what they are...they make
>>>>> sure
>>>>> the levels are hot....just not to the stage of distortion.
>>>>> it's a balancing act, but, hey, who said anything done properly is
>>>>> easy.
>>>>>
>>>
>>>
>>
>>
>
|
|
|
|
| Re: cubase mixing levels [message #75076 is a reply to message #75065] |
Sat, 28 October 2006 19:46 |
Dedric Terry
Messages: 788
Registered: June 2007
|
Senior Member |
|
|
That's why I referred to native summing - most native apps (all but Saw
Plus) are 32-bit floating point.
ProTools is hardware hybrid like Paris so I categorize them differently.
The earlier systems (through Mix+) used a 24-bit data path with no dither,
until they added the dither mixer somewhere in the Mix+ lifecycle, so in
that case you really did lop bits off the bottom when dropping the fader.
The dither engine slightly improved things, but not the lost bits (may have
added a double precision summing section - don't recall specifically).
Still the 24-bit buss was still a big problem.
The only reason PTHD sounds better is they doubled up the processing path to
48bits, but it's the same 24-bit dsp chip family they seem to be stuck with.
Dedric
On 10/28/06 7:37 PM, in article 454403d3@linux, "Aaron Allen"
<know-spam@not_here.dude> wrote:
> If all mix busses are the same, why did the PT III rigs suck out loud so
> badly unless you kicked the master fader down to -15db/-20db?
>
> AA
>
>
> "Dedric Terry" <dterry@keyofd.net> wrote in message
> news:C1684104.4BF6%dterry@keyofd.net...
>> I agree with Martin completely. To add my own general opinion on this long
>> running "sound of summing" debate, that is fast become urban legend:
>>
>> I think it is time to start debunking some of the fables around digital
>> summing. Paris cutting levels and then adding the gain back at the master
>> does one and only one thing: pushes all tracks down by 22dB to make it
>> easier to sum them well below 0dBFS, and then make some of it back before
>> the master so you didn't know what happened. If you are mixing properly
>> in
>> any native DAW, you will in effect do the same thing - lower track levels
>> such that the peak of the summation remains below 0dBFS.
>>
>> In simple math terms:
>>
>> 1- Native DAW: 2+2=4
>> 2- Paris: ((2-1)+(2-1)) + 2 = 4
>>
>> With SX you adjust the gain yourself on the way in, if you like, or better
>> yet, set the levels for the mix at hand as needed, as you go.
>>
>> So why do this in Paris? I am guessing Paris had to convert all audio to
>> 24-bit if sent to the native cpu for native plugins in order to prevent
>> clipping before you even start dropping faders on the mix (unless they
>> somhow converted to 32-bit float on the EDS chip first, which isn't a
>> 32-bit
>> float chip, so that seems unlikely). (No one would want to mix with most
>> faders at -40dB - it would "seem" wrong). 24-bits for any portion of
>> summing (not tracking) is a limitation esp. if tracked audio files are
>> near
>> 0dBFS to begin with - there is no where to go with gain addition, and only
>> subtraction to work with. You can add a lot more -22dB peak audio files
>> to
>> a mix without clipping (with all faders at 0), where only 2 audio files
>> peaking just below 0dBFS will automatically clip. At least that seems to
>> be
>> the reasoning behind it, but imho, only applicable to prevent a potential
>> problem in the DAW itself, not as prescribed digital audio practice.
>>
>> Back to native DAWs: While this approach may seem somehow capable of
>> producing a different sound to a final mix, the only think you gain by
>> lowering tracks by 22dB from the start is lower bit resolution. This is
>> especially true if you happen to record your tracks at lower levels (e.g.
>> -10dB peaks, then your peaks are now at -32dB, which is 1/3 the resolution
>> of 16 bit audio - that isn't insignificant). So, if you combine the
>> flawed
>> advice to record at -10 to -20dB, with the concept of lowering all tracks
>> by
>> 22dB before you start mixing, and you end up removing most of the
>> resolution
>> we work hard for with quality mics, preamps and converters, and mixing as
>> much more noise and quantization error than necessary.
>>
>> As Martin suggested, record just below, or comfortably below clipping (as
>> the source dictates) to maximize your use of bit resolution - e.g. keep
>> audible that ambience or depth of the recording that tails off down at
>> those
>> lower levels, rather than mixing it with mic self-noise, etc.
>>
>> John, what I think the reference you quoted is actually saying (or should
>> be
>> at least) is just that 24-bit digital (with a quality front end and
>> converters) affords a higher signal to noise ratio than analog did, so
>> pushing record levels to widen the gap between peaks and the noise floor
>> isn't as critical. But since quite a few really great mics have a noise
>> floor of around -70 to -74dB, we are still putting noise into half of the
>> bits of that glorious 24-bit range.
>>
>> In terms of signal vs. noise - why record less signal when you can record
>> more?
>>
>> Imho, there are a lot of "famous" engineers out there spouting complete
>> technical rubbish out of lack of true knowledge, or passing along
>> conversational heresay and conjecture. Sadly, engineering is becoming
>> more
>> about letting the gear dictate the recording process than the engineer,
>> and
>> I believe that's what's wrong with music today. Too my people think xyz
>> piece of gear can make a hit, a vibe, or a certain sound just because
>> someone else did it, but few actually use their skills to create the sound
>> by knowing what combinations of gear will help get them there.
>>
>> 9/10 times it isn't about obtaining one piece of gear to get "that sound",
>> but understanding the 1000 different possibilities and knowing how to use
>> any of them.
>>
>> The Nuendo and SX audio engines are identical. They also are identical in
>> summing to Sequoia/Samplitude, and probably Logic, Audition, and DP too
>> (I've tested Nuendo and Sequoia beyond the limits of normal recording to
>> verify this, so this comes from experience and listening, not speculation
>> or
>> internet urban legend).
>>
>> Regards,
>> Dedric
>>
>> On 10/27/06 10:03 PM, in article 4542d488@linux, "alex plasko"
>> <alex.plasko@snet.net> wrote:
>>
>>> hi martin
>>> I think what john is referring to is what started this thread. We are
>>> trying to emulate the way paris handles files at mixdown, not at
>>> recording
>>> files, at mixdown.
>>> chuck said that paris automatically and transparently cuts channel levels
>>> by -22db, and then adds it back automatically when it hits the submix bus
>>> much the way analog consoles do.
>>> what we were toying with is if was possible to emulate that *effect*
>>> with
>>> other daws by cutting channel levels 22db and making it back up at the
>>> output bus.
>>> what we dont know is how cubase/nuendo mix bus handles the files.or
>>> exactly
>>> how paris does it for that matter.
>>> If we can duplicate the way paris handles the mix bus DJ can sleep nights
>>> again .
>>> "Martin Harrington" <lendan@bigpond.net.au> wrote in message
>>> news:4542c966$1@linux...
>>>> Can't tell you anything technically about the Cubase mix bus, (I use
>>>> Nuendo),but I think it's basically the same, but it's a fallacy that if
>>>> you record at lower levels you are protecting the file from clipping.
>>>> What you are doing is not using all the "bits" available to you, and
>>>> therefore start introducing unwanted artifacts into the mix.
>>>> If the "bits" aren't there on the original recording, and the levelis
>>>> cosewuently low at the mix bus, no matter what you do, you can't get
>>>> those
>>>> bits back and the resolution and "size" of your mix has to suffer.
>>>> I record as hot as I can, and use the channel faders to mix, usually
>>>> never
>>>> moving the master fader, although having said that, my mixes for TV/
>>>> Doco
>>>> work are not quite as complicated as most decent size music mixes would
>>>> be.
>>>> --
>>>> Martin Harrington
>>>> www.lendanear-sound.com
>>>> "John" <no@no.com> wrote in message news:45429eda@linux...
>>>>> Martin, so do you know anything about the Cubase mix bus? Do they
>>>>> maybe
>>>>> mean that on mixdown you pull the faders way back but still record hot?
>>>>> Just wondering how the Cubase mix bus behaves.
>>>>>
>>>>> Thanks
>>>>>
>>>>> Martin Harrington wrote:
>>>>>> That's not true, and dont let anyone tell you it is.
>>>>>> You still need to get all levels as optimised as possible, as we all
>>>>>> did
>>>>>> with tape.
>>>>>> Otherwise you are not using all the bits available to you, and noise
>>>>>> will be the end result.
>>>>>> This is why the good/great engineers are what they are...they make
>>>>>> sure
>>>>>> the levels are hot....just not to the stage of distortion.
>>>>>> it's a balancing act, but, hey, who said anything done properly is
>>>>>> easy.
>>>>>>
>>>>
>>>>
>>>
>>>
>>
>
>
|
|
|
|
| Re: cubase mixing levels [message #75080 is a reply to message #75076] |
Sat, 28 October 2006 20:05 |
AlexPlasko
Messages: 211
Registered: September 2006
|
Senior Member |
|
|
ok now i know what you were getting at. i guess i have to follow these
threads closer. skip one or two and i get lost.
thanks dedric
"Dedric Terry" <dterry@keyofd.net> wrote in message
news:C1697220.4C75%dterry@keyofd.net...
> That's why I referred to native summing - most native apps (all but Saw
> Plus) are 32-bit floating point.
>
> ProTools is hardware hybrid like Paris so I categorize them differently.
> The earlier systems (through Mix+) used a 24-bit data path with no dither,
> until they added the dither mixer somewhere in the Mix+ lifecycle, so in
> that case you really did lop bits off the bottom when dropping the fader.
> The dither engine slightly improved things, but not the lost bits (may
> have
> added a double precision summing section - don't recall specifically).
> Still the 24-bit buss was still a big problem.
>
> The only reason PTHD sounds better is they doubled up the processing path
> to
> 48bits, but it's the same 24-bit dsp chip family they seem to be stuck
> with.
>
> Dedric
>
> On 10/28/06 7:37 PM, in article 454403d3@linux, "Aaron Allen"
> <know-spam@not_here.dude> wrote:
>
>> If all mix busses are the same, why did the PT III rigs suck out loud so
>> badly unless you kicked the master fader down to -15db/-20db?
>>
>> AA
>>
>>
>> "Dedric Terry" <dterry@keyofd.net> wrote in message
>> news:C1684104.4BF6%dterry@keyofd.net...
>>> I agree with Martin completely. To add my own general opinion on this
>>> long
>>> running "sound of summing" debate, that is fast become urban legend:
>>>
>>> I think it is time to start debunking some of the fables around digital
>>> summing. Paris cutting levels and then adding the gain back at the
>>> master
>>> does one and only one thing: pushes all tracks down by 22dB to make it
>>> easier to sum them well below 0dBFS, and then make some of it back
>>> before
>>> the master so you didn't know what happened. If you are mixing properly
>>> in
>>> any native DAW, you will in effect do the same thing - lower track
>>> levels
>>> such that the peak of the summation remains below 0dBFS.
>>>
>>> In simple math terms:
>>>
>>> 1- Native DAW: 2+2=4
>>> 2- Paris: ((2-1)+(2-1)) + 2 = 4
>>>
>>> With SX you adjust the gain yourself on the way in, if you like, or
>>> better
>>> yet, set the levels for the mix at hand as needed, as you go.
>>>
>>> So why do this in Paris? I am guessing Paris had to convert all audio
>>> to
>>> 24-bit if sent to the native cpu for native plugins in order to prevent
>>> clipping before you even start dropping faders on the mix (unless they
>>> somhow converted to 32-bit float on the EDS chip first, which isn't a
>>> 32-bit
>>> float chip, so that seems unlikely). (No one would want to mix with
>>> most
>>> faders at -40dB - it would "seem" wrong). 24-bits for any portion of
>>> summing (not tracking) is a limitation esp. if tracked audio files are
>>> near
>>> 0dBFS to begin with - there is no where to go with gain addition, and
>>> only
>>> subtraction to work with. You can add a lot more -22dB peak audio files
>>> to
>>> a mix without clipping (with all faders at 0), where only 2 audio files
>>> peaking just below 0dBFS will automatically clip. At least that seems
>>> to
>>> be
>>> the reasoning behind it, but imho, only applicable to prevent a
>>> potential
>>> problem in the DAW itself, not as prescribed digital audio practice.
>>>
>>> Back to native DAWs: While this approach may seem somehow capable of
>>> producing a different sound to a final mix, the only think you gain by
>>> lowering tracks by 22dB from the start is lower bit resolution. This is
>>> especially true if you happen to record your tracks at lower levels
>>> (e.g.
>>> -10dB peaks, then your peaks are now at -32dB, which is 1/3 the
>>> resolution
>>> of 16 bit audio - that isn't insignificant). So, if you combine the
>>> flawed
>>> advice to record at -10 to -20dB, with the concept of lowering all
>>> tracks
>>> by
>>> 22dB before you start mixing, and you end up removing most of the
>>> resolution
>>> we work hard for with quality mics, preamps and converters, and mixing
>>> as
>>> much more noise and quantization error than necessary.
>>>
>>> As Martin suggested, record just below, or comfortably below clipping
>>> (as
>>> the source dictates) to maximize your use of bit resolution - e.g. keep
>>> audible that ambience or depth of the recording that tails off down at
>>> those
>>> lower levels, rather than mixing it with mic self-noise, etc.
>>>
>>> John, what I think the reference you quoted is actually saying (or
>>> should
>>> be
>>> at least) is just that 24-bit digital (with a quality front end and
>>> converters) affords a higher signal to noise ratio than analog did, so
>>> pushing record levels to widen the gap between peaks and the noise floor
>>> isn't as critical. But since quite a few really great mics have a noise
>>> floor of around -70 to -74dB, we are still putting noise into half of
>>> the
>>> bits of that glorious 24-bit range.
>>>
>>> In terms of signal vs. noise - why record less signal when you can
>>> record
>>> more?
>>>
>>> Imho, there are a lot of "famous" engineers out there spouting complete
>>> technical rubbish out of lack of true knowledge, or passing along
>>> conversational heresay and conjecture. Sadly, engineering is becoming
>>> more
>>> about letting the gear dictate the recording process than the engineer,
>>> and
>>> I believe that's what's wrong with music today. Too my people think xyz
>>> piece of gear can make a hit, a vibe, or a certain sound just because
>>> someone else did it, but few actually use their skills to create the
>>> sound
>>> by knowing what combinations of gear will help get them there.
>>>
>>> 9/10 times it isn't about obtaining one piece of gear to get "that
>>> sound",
>>> but understanding the 1000 different possibilities and knowing how to
>>> use
>>> any of them.
>>>
>>> The Nuendo and SX audio engines are identical. They also are identical
>>> in
>>> summing to Sequoia/Samplitude, and probably Logic, Audition, and DP too
>>> (I've tested Nuendo and Sequoia beyond the limits of normal recording to
>>> verify this, so this comes from experience and listening, not
>>> speculation
>>> or
>>> internet urban legend).
>>>
>>> Regards,
>>> Dedric
>>>
>>> On 10/27/06 10:03 PM, in article 4542d488@linux, "alex plasko"
>>> <alex.plasko@snet.net> wrote:
>>>
>>>> hi martin
>>>> I think what john is referring to is what started this thread. We are
>>>> trying to emulate the way paris handles files at mixdown, not at
>>>> recording
>>>> files, at mixdown.
>>>> chuck said that paris automatically and transparently cuts channel
>>>> levels
>>>> by -22db, and then adds it back automatically when it hits the submix
>>>> bus
>>>> much the way analog consoles do.
>>>> what we were toying with is if was possible to emulate that *effect*
>>>> with
>>>> other daws by cutting channel levels 22db and making it back up at the
>>>> output bus.
>>>> what we dont know is how cubase/nuendo mix bus handles the files.or
>>>> exactly
>>>> how paris does it for that matter.
>>>> If we can duplicate the way paris handles the mix bus DJ can sleep
>>>> nights
>>>> again .
>>>> "Martin Harrington" <lendan@bigpond.net.au> wrote in message
>>>> news:4542c966$1@linux...
>>>>> Can't tell you anything technically about the Cubase mix bus, (I use
>>>>> Nuendo),but I think it's basically the same, but it's a fallacy that
>>>>> if
>>>>> you record at lower levels you are protecting the file from clipping.
>>>>> What you are doing is not using all the "bits" available to you, and
>>>>> therefore start introducing unwanted artifacts into the mix.
>>>>> If the "bits" aren't there on the original recording, and the levelis
>>>>> cosewuently low at the mix bus, no matter what you do, you can't get
>>>>> those
>>>>> bits back and the resolution and "size" of your mix has to suffer.
>>>>> I record as hot as I can, and use the channel faders to mix, usually
>>>>> never
>>>>> moving the master fader, although having said that, my mixes for TV/
>>>>> Doco
>>>>> work are not quite as complicated as most decent size music mixes
>>>>> would
>>>>> be.
>>>>> --
>>>>> Martin Harrington
>>>>> www.lendanear-sound.com
>>>>> "John" <no@no.com> wrote in message news:45429eda@linux...
>>>>>> Martin, so do you know anything about the Cubase mix bus? Do they
>>>>>> maybe
>>>>>> mean that on mixdown you pull the faders way back but still record
>>>>>> hot?
>>>>>> Just wondering how the Cubase mix bus behaves.
>>>>>>
>>>>>> Thanks
>>>>>>
>>>>>> Martin Harrington wrote:
>>>>>>> That's not true, and dont let anyone tell you it is.
>>>>>>> You still need to get all levels as optimised as possible, as we all
>>>>>>> did
>>>>>>> with tape.
>>>>>>> Otherwise you are not using all the bits available to you, and noise
>>>>>>> will be the end result.
>>>>>>> This is why the good/great engineers are what they are...they make
>>>>>>> sure
>>>>>>> the levels are hot....just not to the stage of distortion.
>>>>>>> it's a balancing act, but, hey, who said anything done properly is
>>>>>>> easy.
>>>>>>>
>>>>>
>>>>>
>>>>
>>>>
>>>
>>
>>
>
|
|
|
|
| Re: cubase mixing levels [message #75084 is a reply to message #74844] |
Sat, 28 October 2006 22:53 |
Dedric Terry
Messages: 788
Registered: June 2007
|
Senior Member |
|
|
Amplitude quantization is linear in the sense that we are subdividing
voltage in voltage steps - e.g. if the full voltage range is 10V (+/- 5V),
then with 16 bits, we have 65,536 divisions, each representing a 153uV step.
As far as dynamic range, if you are recording at 24 bit, then -48dB would be
technically represented by the lower 16 bits. 8 bits gives us 48dB of
dynamic range, but -48dB within a 144dB dynamic range would fall in the
lower 16 bits, since 0dBFS is the maximum power ratio for any bit depth.
You would still be recording/filling 24-bits, but the upper bits would be
0's.
Software encoding of amplitude would represent this a bit differently, but
the theory in terms of how bit depth relates to representing the audio level
we refer to is the same.
Dedric
On 10/28/06 6:01 PM, in article 4543e130$1@linux, "Neil" <OIUOIU@OIU.com>
wrote:
>
> "Dedric Terry" <dedric@echomg.com> wrote:
>
>> Yes with amplitude you get greater resolution with higher bit
>> depths (more levels to quantize each sample).
>
> Yes, that's what I meant, and it's linear, right? In other
> words, if a sine wave is full-on 0db flat-out at peaks, then
> (in a simplistic example) 12 bits are taken up in half that
> amplitude, and 12 bits are taken up in the remaining half, or 6
> bits are taken up in a quarter of it, and so on, right?
>
>> This isn't the same as the dynamic range though.
>
> Ooooooooh, ok, now I'm confused again.
>
> If you're recording at a low level... let's say peaks of -40
> or -50 db on a given track, chances are you're only using 12 or
> so bits of that file capacity even though you may
> be "recording" at 24-bits, yes? Or are you saying becasue db is
> log, not linear, you might actually be only using 6 or so of
> your available bits if you're recording at those low peaks of
> -40 or -50?
>
> Neil
|
|
|
|
|
|
| Re: cubase mixing levels [message #75130 is a reply to message #75008] |
Mon, 30 October 2006 05:33 |
animix
Messages: 356
Registered: September 2006
|
Senior Member |
|
|
I knew that a cowbell would somehow ind it's way into this.
"Neil" <OIUOIU@OIU.com> wrote in message news:45436d90$1@linux...
>
> Dedric Terry <dterry@keyofd.net> wrote:
>
> >I agree with Martin completely. To add my own general opinion
> >on this long running "sound of summing" debate,
>
> Great post, Dedric! BTW, "The Sound of Summing" - wasn't that a
> song by Summon & Nullfunkel?
>
> Anyway, a couple of things I wanted to bring up. You said:
>
> >In simple math terms:
> >1- Native DAW: 2+2=4
> >2- Paris: ((2-1)+(2-1)) + 2 = 4
> >With SX you adjust the gain yourself on the way in, if you like, or
better
> >yet, set the levels for the mix at hand as needed, as you go.
>
> I just wanted to point out that if anyone thought I meant to
> adjust your levels "on the way in" (i.e.: using the Channel
> input trim control), when I had said to default all your
> channels to -6, for example, as a starting point - that I did
> NOT, in fact, mean the input trims!!! I meant the regular
> ol' "please make it louder or softer" control. :D
>
>
> >So why do this in Paris? I am guessing Paris had to convert all audio to
> >24-bit if sent to the native cpu for native plugins in order to prevent
> >clipping before you even start dropping faders on the mix (unless they
> >somhow converted to 32-bit float on the EDS chip first, which isn't a
32-bit
> >float chip, so that seems unlikely). (No one would want to mix with most
> >faders at -40dB - it would "seem" wrong). 24-bits for any portion of
> >summing (not tracking) is a limitation esp. if tracked audio files are
near
> >0dBFS to begin with - there is no where to go with gain addition, and
only
> >subtraction to work with. You can add a lot more -22dB peak audio files
> to
> >a mix without clipping (with all faders at 0), where only 2 audio files
> >peaking just below 0dBFS will automatically clip. At least that seems to
> be
> >the reasoning behind it, but imho, only applicable to prevent a potential
> >problem in the DAW itself, not as prescribed digital audio practice.
> >
> >Back to native DAWs: While this approach may seem somehow capable of
> >producing a different sound to a final mix, the only think you gain by
> >lowering tracks by 22dB from the start is lower bit resolution. This is
> >especially true if you happen to record your tracks at lower levels (e.g.
> >-10dB peaks, then your peaks are now at -32dB, which is 1/3 the
resolution
> >of 16 bit audio - that isn't insignificant). So, if you combine the
flawed
> >advice to record at -10 to -20dB, with the concept of lowering all tracks
> by
> >22dB before you start mixing, and you end up removing most of the
resolution
> >we work hard for with quality mics, preamps and converters, and mixing as
> >much more noise and quantization error than necessary
>
> Which is why I suggest starting at -6... maybe -10 if you're
> going to be loading up a buttload of tracks. Think about it
> with an analog analogy again, gang: If your analog mixing
> console goes up to +10 on each channel, would you start a mix
> with every fucking fader maxed out at +10??? Hit "play", and
> how would THAT summing buss sound right about then? No, to
> start out with, you'd bring all the faders down to 0, or -5,
> or -10, or whatever the your comfortable starting point was,
> knowing the console & how much headroom it's got, etc., right?
> Pushing everything up to +10 to start off with would be just
> simply way too much, yes?
>
> So why are people not willing to get their heads around the
> fact that in digital anything past "0" is "way too much"?
> The only reason your DAW has +6 or +8 on any given channel is
> in case you fuck up & record the cowbell track on your band's
> cover version of "Mississippi Queen" at peak levels of -17db
> ... "0" gain on the channel level just wouldn't cut it at that
> stage, the cowbell just wouldn't be audible enough to drive
> that tune. :D
>
> So, just like in the analog console, where 40+ channels
> recorded to needle-bending levels on overbiased tape machines,
> and every channel set to +10 at the start of your mixdown
> session would sound like crap; so does 40+ channels recorded to
> nice hot levels, barely missing overs by a tenth of a db,
> with every channel set to "0" result in a similar thing.
>
> Neil
>
|
|
|
|
| Re: cubase mixing levels [message #75217 is a reply to message #75086] |
Tue, 31 October 2006 08:31 |
fernando
Messages: 15
Registered: October 2006
|
Junior Member |
|
|
Do you know, my E-MU 1212M and new 1616M PCI cards has the
same converters as PTHD, A/D convertor is AK5394A and D/A
convertor is CS4398, sounds real good for not colored hi-end stuff. Very
cheap too, $199 with bundle of SOANAR LE, Cubase LE, Ableton Live Lite 4,
Wavelab Lite and other cool stuff like Proteus X LE. But I think 1212M is
now not supported, sounds familiar, yes?
"Neil" <OIUOIU@OIU.com> wrote:
>
>Dedric Terry <dterry@keyofd.net> wrote:
>>That's why I referred to native summing - most native apps (all but Saw
>>Plus) are 32-bit floating point.
>>
>>ProTools is hardware hybrid like Paris so I categorize them differently.
>>The earlier systems (through Mix+) used a 24-bit data path with no dither,
>>until they added the dither mixer somewhere in the Mix+ lifecycle, so in
>>that case you really did lop bits off the bottom when dropping the fader.
>>The dither engine slightly improved things, but not the lost bits (may
have
>>added a double precision summing section - don't recall specifically).
>>Still the 24-bit buss was still a big problem.
>
>Plus, those 888 convertors sucked hind tit - that sure didn't
>help matters, summing issues or no.
>
>>The only reason PTHD sounds better is they doubled up the processing path
>to 48bits
>
>And the newer Digi convertors really do sound good - start with
>crap, end with crap - highly polished crap, but crap
>nonetheless. Start with something good - it's up to the user to
>fuck it up from there! LOL
>
>Neil
|
|
|
|
Goto Forum:
Current Time: Fri Nov 29 10:20:59 PST 2024
Total time taken to generate the page: 0.03150 seconds
|
The PARIS Forums
Search
Help
Members
Register
Login
Home
