17 Apr
2004
17 Apr
'04
3:39 p.m.
Date: Fri, 16 Apr 2004 09:43:00 -0700 (PDT)
From: R Parker <rtp405(a)yahoo.com>
Subject: [linux-audio-user] sndfile-info and lots of questions
To: linux-audio-user(a)music.columbia.edu
Message-ID: <20040416164300.90068.qmail(a)web20916.mail.yahoo.com>
Content-Type: text/plain; charset=us-ascii
Hi,
As I've gone through my most challanging mastering
project, I've developed more questions than answers.
Hope it all turned out right in the end!
Only just caught up with the last email. I reackon Jamin could easily
sort out the occasional uncontrollable kick drum. Perhaps in a two pass
process, once to sort out that, then another pass for more general
mastering.
There's been reference, on this list, to
documentation
that explains file formats-- I don't recall the
document title or where to find it. Ultimately, I need
a resouce that explains things like; the number of
available samples for the different bit depths (16bit
range from -ABC to +XYZ, 24bit...), DC offset is, peak
amplitude is, RMS is, etc.
I need to know if samples are syncronous with decibel
level, is maximum samples equal to 0db?
Yes. Though 0dbfs is perhaps a better way to put it.
Samples in 16bit= 2^16 Samples = -96dbfs range.
Samples in 24bit= 2^24 Samples = -144dbfs range.
Samples in 32bit float = I'm not sure if you count the Mantissa or not =
Shockingly massive dynamic range. (Almost exactly :) ).
Of course another challange is tools like sndfile-info
assume that a file exists. This is not always the case
and in my situation it's almost never true. I return
JAMin output to an Ardour return bus and don't produce
a file until the return bus is exported. Printing a
track to the file system and then analyzing it is no
way to save time.
I guess the only way some of those figures can be found is for the file
to be printed. Ie, you can only tell what the peak sample is if you have
compared it to every other sample in the file. To do it in real time,
you would have to have 'Peak Sample so far', 'RMS so far' etc, and reset
the figures manually when you started playback again. Not to say such a
facility would not be useful. It would be nice if Jamin had a statistics
page that showed you how many overs there had been, how many of them
were consecutive etc.
There is a new program called'Jackbitscope'
http://www.ecs.soton.ac.uk/~njl98r/code/ladspa/bitscope/
this has displays for highest and lowest sample so far. (And lots of
other useful looking things I don't really understand. )
Anyway, I appreciate all the responses to my past
questions and am hopeful that someone can look at the
current mumbo jumbo and prescribe some effective
medications; coffee, sleep, black bear gallbladders,
urls to useful documents, etc.
Ground lion's teeth are effective in 90% of these cases in my
experience. If not.....
"0dbfs+ levels in digital mastering"
http://www.studio-systems.com/audiofeatures/SepOct2002/Audio%20Article/0DBF…
It's a couple of years old but still interesting, in that it covers what
can happen when very hot digital levels hit the analog world.
Also, the rec.audio.pro faq is chock full of useful stuff...
http://www.faqs.org/faqs/AudioFAQ/pro-audio-faq/
26 Apr
26 Apr
12:58 p.m.
New subject: [linux-audio-user] sndfile-info and lots
Hi,
--- philicorda <philicorda(a)ntlworld.com> wrote:
http://www.ecs.soton.ac.uk/~njl98r/code/ladspa/bitscope/
Hope it all turned out right in the end!
I'm pretty confidant. As you may recall the order was
to achieve greater volume across a 13 song album. I
definitely accomplished that.
I installed Bitscope. Its numerical report for decibal
level of "Largest sample" combined with metering in
all the stages helped push the levels towards 0db.
I've always been nervous about hot levels but I think
we've got enough usable tools to confidantly do this
type of work.
After completing the first version, I referenced
against some "million dollar" masters and discovered
that my results are louder.
This was a useful discovery because it allowed me to
back the limiting off of several songs and minimize
the "munge" affect.
My overall review of mastering with Linux tools is
that we have what's needed to do acceptable work. All
of the tools have the expected bugs. No big deal. The
real upside to usability is that jack enables us to
configure routing scenarios that are perfect for this
type of work.
I used an Ardour session with 26 stereo tracks as the
source for JAMin. Thirteen of those tracks were for
existing masters and the other thirteen for the
unmastered control room mixes. At the hardware mixer,
I was able to use the mute button to switch between
the two sources and have the perfect A to B
comparisons.
When Ardour stabilizes and all its features become
usable, it and JAMin create an ideal setup. One of the
guys, Uwe, has written a LADSP module to automate the
scene changes in JAMin--I think it can be seen in
Rezound. Anyway, as things flush out I'm sure we'll
produce some documents describing how to setup an
ideal mastering environment.
Another studio owner friend stopped by and saw what I
was doing. He was impressed and said he's gonna buy an
x86 box and check out the linux stuff.
I've still got plenty to learn but with all the help
from everyone on this list I've been able to produce
an accepted product. Thanks much!
ron
There's
been reference, on this list, to
documentation
that explains file formats-- I don't recall
the
document title or where to find it. Ultimately, I
need
a resouce that explains things like; the number of
available samples for the different bit depths
(16bit
range from -ABC to +XYZ, 24bit...), DC offset is,
peak
amplitude is, RMS is, etc.
I need to know if samples are syncronous with
decibel
level, is maximum samples equal to 0db?
Yes. Though 0dbfs is perhaps a better way to put it.
Samples in 16bit= 2^16 Samples = -96dbfs range.
Samples in 24bit= 2^24 Samples = -144dbfs range.
Samples in 32bit float = I'm not sure if you count
the Mantissa or not =
Shockingly massive dynamic range. (Almost exactly :)
).
Of course another challange is tools like
sndfile-info
assume that a file exists. This is not always the
case
and in my situation it's almost never true. I
return
JAMin output to an Ardour return bus and don't
produce
a file until the return bus is exported. Printing
a
track to the file system and then analyzing it is
no
way to save time.
I guess the only way some of those figures can be
found is for the file
to be printed. Ie, you can only tell what the peak
sample is if you have
compared it to every other sample in the file. To do
it in real time,
you would have to have 'Peak Sample so far', 'RMS so
far' etc, and reset
the figures manually when you started playback
again. Not to say such a
facility would not be useful. It would be nice if
Jamin had a statistics
page that showed you how many overs there had been,
how many of them
were consecutive etc.
There is a new program called'Jackbitscope'
this has displays for highest and lowest sample so
far. (And lots of
other useful looking things I don't really
understand. )
http://www.studio-systems.com/audiofeatures/SepOct2002/Audio%20Article/0DBF…
Anyway, I appreciate all the responses to my past
questions and am hopeful that someone can look at
the
current mumbo jumbo and prescribe some effective
medications; coffee, sleep, black bear
gallbladders,
urls to useful documents, etc.
Ground lion's teeth are effective in 90% of these
cases in my
experience. If not.....
"0dbfs+ levels in digital mastering"
It's a couple of years old but still interesting,
in
that it covers what
can happen when very hot digital levels hit the
analog world.
Also, the rec.audio.pro faq is chock full of useful
stuff...
http://www.faqs.org/faqs/AudioFAQ/pro-audio-faq/
__________________________________
Do you Yahoo!?
Yahoo! Photos: High-quality 4x6 digital prints for 25�
http://photos.yahoo.com/ph/print_splash
1:11 p.m.
New subject: [linux-audio-user] sndfile-info and lots
On Mon, 2004-04-26 at 11:51, R Parker wrote:
When Ardour stabilizes and all its features become
usable, it and JAMin create an ideal setup. One of the
guys, Uwe, has written a LADSP module to automate the
scene changes in JAMin--I think it can be seen in
Rezound. Anyway, as things flush out I'm sure we'll
produce some documents describing how to setup an
ideal mastering environment.
Speaking of which - is this patch for the latest CVS? If not, can I get
a copy of it to see what needs to be done?
Jan
7701
days inactive
7710
days old
linux-audio-user@lists.linuxaudio.org
2 comments
3 participants
participants (3)
-
Jan Depner -
philicorda -
R Parker