Ok I know I'm opening up a can of worms here that will likely degenerate into a
flame war, but hopefully I'll be able to extract some useful information from
the thread before that happens.

So can anyone give me a rundown, as elaborate as possible, on the differences
between left and right justified 24bit files, and 24bit 'packed' files, and the
relevence of those differences? I already know that left justified and right
justified 24bit words are actually 32bit words in which the 24bit data is
either 'pushed' to the right or the left like this:

("X" is the audio data, "0" is the unused portion of the 32-bit data)

Left-justified: XXXXXXXXXXXXXXXXXXXXXXXX00000000
Right-justified: 00000000XXXXXXXXXXXXXXXXXXXXXXXX


But what does a packed 24bit word look like, and what does the 'packing'
process entail? Which format is best to use? What sort of quality degradation
occurs when a right justified file is played back on a soundcard configured for
left justification? Is there any way to determine which a soundcard interface
uses if it doesn't say so in the documentation or driver control panel?

Thanks in advance for any info, but do us all a favor and sit back and read
along with me if you don't know much more about this than I do, because we'll
all benefit most if we only hear from people who really know what they're
talking about in regards to this topic!

In article 3hscg.26467$Fl1.24321@edtnps89, Chevdo wrote:

So can anyone give me a rundown, as elaborate as possible, on the differences
between left and right justified 24bit files, and 24bit 'packed' files, and the
relevence of those differences? I already know that left justified and right
justified 24bit words are actually 32bit words in which the 24bit data is
either 'pushed' to the right or the left like this:

("X" is the audio data, "0" is the unused portion of the 32-bit data)

Left-justified: XXXXXXXXXXXXXXXXXXXXXXXX00000000
Right-justified: 00000000XXXXXXXXXXXXXXXXXXXXXXXX

Correct.

But what does a packed 24bit word look like, and what does the 'packing'
process entail?

The packing process involves splitting the 24 bit stream up into 32 bit
chunks. The first word has the first 3 bytes of the first sample, plus
the first byte of the second. The next word has the last two bytes of
the second plus the first two bytes of the third.

Nothing is padded out, so the file size is smaller. But it requires a
few more machine cycles to strip the thing into individual samples.

Which format is best to use? What sort of quality degradation
occurs when a right justified file is played back on a soundcard configured for
left justification? Is there any way to determine which a soundcard interface
uses if it doesn't say so in the documentation or driver control panel?

The soundcard doesn't care. The soundcard just gets bits. If you play a
right-justfied file into a thing that expects a left-justified one, or
vice-versa, it will chop the leading or trailing byte off and the result
will be a distorted mess. But the software should know all about this and
deal with it properly. If it doesn't, you'll know pretty quickly.

There isn't a damn bit of difference between using the three file formats,
other than the packed file taking a few extra cycles to massage.

Thanks in advance for any info, but do us all a favor and sit back and read
along with me if you don't know much more about this than I do, because we'll
all benefit most if we only hear from people who really know what they're
talking about in regards to this topic!

You worry too much.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

In article , says...

In article 3hscg.26467$Fl1.24321@edtnps89, Chevdo wrote:

So can anyone give me a rundown, as elaborate as possible, on the differences
between left and right justified 24bit files, and 24bit 'packed' files, and
the
relevence of those differences? I already know that left justified and right
justified 24bit words are actually 32bit words in which the 24bit data is
either 'pushed' to the right or the left like this:

("X" is the audio data, "0" is the unused portion of the 32-bit data)

Left-justified: XXXXXXXXXXXXXXXXXXXXXXXX00000000
Right-justified: 00000000XXXXXXXXXXXXXXXXXXXXXXXX

Correct.

Of course. I wasn't asking for validation for what I know.


But what does a packed 24bit word look like, and what does the 'packing'
process entail?

The packing process involves splitting the 24 bit stream up into 32 bit
chunks. The first word has the first 3 bytes of the first sample, plus
the first byte of the second. The next word has the last two bytes of
the second plus the first two bytes of the third.

Nothing is padded out, so the file size is smaller. But it requires a
few more machine cycles to strip the thing into individual samples.


Ahh that makes sense, thanks for explaining it.



The soundcard doesn't care. The soundcard just gets bits. If you play a
right-justfied file into a thing that expects a left-justified one, or
vice-versa, it will chop the leading or trailing byte off and the result
will be a distorted mess. But the software should know all about this and
deal with it properly. If it doesn't, you'll know pretty quickly.

That's not what the CardDeluxe FAQ where I obtained what little information
about this subject I know indicates. According to the FAQ,

"if you are setup for 24-bit left-justified, and the audio application is
trying to work in right-justified, you will not be recording or playing the
data correctly. You will need to check your audio application to see how it
deals with 24-bit unpacked data, so you can set our driver accordingly. Your
audio application may also be able to make this change within its preferences
setup. Most applications, at this time, use left-justified, which is what the
CardDeluxe is set to use by default."

So according to the CardDeluxe FAQ you will not know 'pretty quickly' if their
is a mismatch between the soundcard interface and software regarding
justification. According to the FAQ, "You will need to check your audio
application to see how it deals with 24-bit unpacked data". Unfortunately I
don't use a CardDeluxe anymore, and the M-audio R-BUS card I use doesn't have a
setting to switch between left or right justified unpacked data. Furthermore,
it appears that my preferred DAW application, SawPro, records 24bit packed
data, which I never would have known if I hadn't imported my SawPro session
into SawStudio which indicated that the recordings I did in SawPro are packed.
I'd rather not use packed, I'd rather have full control over both software and
hardware to use left-justified unpacked, but that doesn't seem to be an option
with the R-BUS drivers nor with SawPro or SawStudio.




There isn't a damn bit of difference between using the three file formats,
other than the packed file taking a few extra cycles to massage.


It'd be nice if I could take your word for that, but I trust the CardDeluxe
engineers a bit more than I trust you. But I do appreciate the response,
because your explaination of what the packed words were all about sounds right
and was informative to me. I hope someone else responds to add more
informative comments to the subject..

In article AYtcg.26572$Fl1.3511@edtnps89, says...


So according to the CardDeluxe FAQ you will not know 'pretty quickly' if their
is a mismatch

yikes, I meant 'there', not 'their'. Very uncommon mistake for me to make, I
swear!

In article AYtcg.26572$Fl1.3511@edtnps89, Chevdo wrote:

The soundcard doesn't care. The soundcard just gets bits. If you play a
right-justfied file into a thing that expects a left-justified one, or
vice-versa, it will chop the leading or trailing byte off and the result
will be a distorted mess. But the software should know all about this and
deal with it properly. If it doesn't, you'll know pretty quickly.

That's not what the CardDeluxe FAQ where I obtained what little information
about this subject I know indicates. According to the FAQ,

"if you are setup for 24-bit left-justified, and the audio application is
trying to work in right-justified, you will not be recording or playing the
data correctly. You will need to check your audio application to see how it
deals with 24-bit unpacked data, so you can set our driver accordingly. Your
audio application may also be able to make this change within its preferences
setup. Most applications, at this time, use left-justified, which is what the
CardDeluxe is set to use by default."

So according to the CardDeluxe FAQ you will not know 'pretty quickly' if their
is a mismatch between the soundcard interface and software regarding
justification. According to the FAQ, "You will need to check your audio
application to see how it deals with 24-bit unpacked data".

If there is a mismatch, you will either be truncating the MSB or the LSB.
If you truncate the MSB, the effect is painfully obvious to the point where
you won't be able to understand the words. If you truncate the LSB, the
effect is not quite as dramatic, but you'll still hear reverb tails sounding
very unnatural at low levels.

setting to switch between left or right justified unpacked data. Furthermore,
it appears that my preferred DAW application, SawPro, records 24bit packed
data, which I never would have known if I hadn't imported my SawPro session
into SawStudio which indicated that the recordings I did in SawPro are packed.
I'd rather not use packed, I'd rather have full control over both software and
hardware to use left-justified unpacked, but that doesn't seem to be an option
with the R-BUS drivers nor with SawPro or SawStudio.

Okay, you have two different things you are talking about here. First is the
file format, which is one thing. The second is the format of the data sent
to the card. These are independant of one another. You can have the card set
one way and use a file with a different format.

The application knows how the file is configured, because the file header
tells it.

The application knows how the soundcard is configured either because the
driver tells it, or you set it yourself in a menu.

There isn't a damn bit of difference between using the three file formats,
other than the packed file taking a few extra cycles to massage.


It'd be nice if I could take your word for that, but I trust the CardDeluxe
engineers a bit more than I trust you. But I do appreciate the response,
because your explaination of what the packed words were all about sounds right
and was informative to me. I hope someone else responds to add more
informative comments to the subject..

There IS a difference between the packing in the data path between the
application and the driver. That's what the CardDeluxe guys are talking
about. This is not necessarily the same as the format of the file.

The file can be ANY format. As long as the application knows what format
it is, and what format the card is, translation should be a non-issue.
The problems described occur when the application doesn't know what format
the card is.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

In article AYtcg.26572$Fl1.3511@edtnps89, says...

The soundcard doesn't care. The soundcard just gets bits. If you play a
right-justfied file into a thing that expects a left-justified one, or
vice-versa, it will chop the leading or trailing byte off and the result
will be a distorted mess. But the software should know all about this and
deal with it properly. If it doesn't, you'll know pretty quickly.

That's not what the CardDeluxe FAQ where I obtained what little information
about this subject I know indicates. According to the FAQ,


ok well I found this from the Sawstudio forum, from the author of the softwa

"It (unpacked format) can sometimes also possibly improve performance due to
the fact that our current Pc's access data in DWORD chunks and the packed
format requires multiple register shifts to compact and expand the data into
the odd size 3 byte format, which translates into more internal manipulation to
access each sample.

In most situations the difference may not be noticable.

For compatibility with other 24 bit wav files in most other software, you
should use the default packed format."


Which is strange, since the CardDeluxe people seem to think left justified
unpacked should be the default. And I have no idea what M-Audio thinks should
be the default for their R-BUS cards...

In article , says...

If there is a mismatch, you will either be truncating the MSB or the LSB.
If you truncate the MSB, the effect is painfully obvious to the point where
you won't be able to understand the words. If you truncate the LSB, the
effect is not quite as dramatic, but you'll still hear reverb tails sounding
very unnatural at low levels.


ok, I don't know how good I am at noticing unnatural reverb tails at low
levels, though.

But

Okay, you have two different things you are talking about here. First is the
file format, which is one thing. The second is the format of the data sent
to the card. These are independant of one another. You can have the card set
one way and use a file with a different format.

I'm not sure what you're referring to, but the CardDeluxe FAQ makes it pretty
clear that for optiumum performance, both the software and the hardware need to
be set to output and input the same format.


The application knows how the soundcard is configured either because the
driver tells it, or you set it yourself in a menu.

Well I can't set it, so I guess I have to hope the driver tells both the
software and the hardware to use the same format. I wish I knew, and I also
wish I could set everything to use unpacked, because it's clear that packed
uses more processing, and if it means I can use a few more EQ plugins or
whatever before my CPU craps out by using unpacked rather than packed, then I
want to be able to use unpacked!


There IS a difference between the packing in the data path between the
application and the driver. That's what the CardDeluxe guys are talking
about. This is not necessarily the same as the format of the file.

The file can be ANY format. As long as the application knows what format
it is, and what format the card is, translation should be a non-issue.
The problems described occur when the application doesn't know what format
the card is.
-

I can't really figure out what your fixation on file format is all about. The
recorded audio will get saved in a file in whatever format the audio was set to
record as. The software will then of course play it back as that format, but
whether or not the software uses the driver, or can use the driver, to switch
to that format upon playback if the device is not already set to playback that
format, is apparently a matter of whether or not the programmer of the
software and driver knew what they were doing. I would say that with the
equipment I'm using, M-audio's R-BUS and SawPro, that they did not, since they
did not provide any means of configuring these parameters. It appears to me
that I am stuck using packed 24bit format because the driver and the software
were both made to revert to that format as the most compatible. And that sucks
because I wish I could get the better performance by using unpacked format.

1) Interface to the card.
Unless you program your own driver routines, this is not an issue.
Today you simply use one of the driver formats (ASIO or WDM) and be
done with it. If you program your own driver routines, some people do
that in Linux as example, it is of course different. I might add that
writing sound card drivers is not for the feeble minded.

2) The file format - packed or not
The performance difference is not noticeable on a modern computer. What
matters is if you want to export the files between different programs,
then a standard format such as .wav is preferrable.

Things used to be different as you had to tweak and twiddle a lot at a
low level. Most of these things are gone today as computers has
progressed.

Gunnar

Rather common otherwise... Like confusing "your" and "you're"...


And "it's" or "its"...

On Tue, 23 May 2006 21:13:15 +1200, "amp_noob"
wrote:


Rather common otherwise... Like confusing "your" and "you're"...


And "it's" or "its"...


You've all missed the worst one - "alot" for "a lot".

d

--
Pearce Consulting
http://www.pearce.uk.com

"Chevdo" wrote in message
news:AYtcg.26572$Fl1.3511@edtnps89
In article ,
says...

The soundcard doesn't care. The soundcard just gets
bits. If you play a right-justfied file into a thing
that expects a left-justified one, or vice-versa, it
will chop the leading or trailing byte off and the
result will be a distorted mess. But the software
should know all about this and deal with it properly.
If it doesn't, you'll know pretty quickly.

That's not what the CardDeluxe FAQ where I obtained what
little information about this subject I know indicates.

Wrong. The CardDeluxe says a similar thing, just with a little different
spin.

According to the FAQ,

"if you are setup for 24-bit left-justified, and the
audio application is trying to work in right-justified,
you will not be recording or playing the data correctly.
You will need to check your audio application to see how
it deals with 24-bit unpacked data, so you can set our
driver accordingly. Your audio application may also be
able to make this change within its preferences setup.
Most applications, at this time, use left-justified,
which is what the CardDeluxe is set to use by default."

First off, this paragraph is in fact not all that well-written. It
contradicts itself. First it says that you will have to set their driver
accordingly, and then it says that the application may be able to adjust the
setting for itself.

Other than contradicting itself, the Card Deluxe FAQ does not disagree with
Scott's statement in any way.

So according to the CardDeluxe FAQ you will not know
'pretty quickly' if their is a mismatch between the
soundcard interface and software regarding justification.

Actually, the CardDeluxe FAQ portion you quoted does not comment on whether
or not you will notice the difference when you listen. It does fail to point
out the fact that in many cases, the mismatch is painfully obvious. We're
talking about changing *any* audio signal into a grotesque-sounding noise.

Chevdo, I can tell you've never had the *pleasure* of hearing this kind of
mismatch or you wouldn't go on like this. So, this is about your
inexperience, not a proven conflict between the CardDeluxe FAQ and Scott.

According to the FAQ, "You will need to check your audio
application to see how it deals with 24-bit unpacked
data".

The FAQ then contracts itself by saying that the application might handle
the sitaution automatically so that you won't have to check anything.

Unfortunately I don't use a CardDeluxe anymore,
and the M-audio R-BUS card I use doesn't have a setting
to switch between left or right justified unpacked data.

However, the application might be able to correct the situation all by
itself. Any necessary transformation is pretty trivial to program.

Furthermore, it appears that my preferred DAW
application, SawPro, records 24bit packed data, which I
never would have known if I hadn't imported my SawPro
session into SawStudio which indicated that the
recordings I did in SawPro are packed.

Really good software might make this level of detail transparent to the
user.

I'd rather not use
packed, I'd rather have full control over both software
and hardware to use left-justified unpacked, but that
doesn't seem to be an option with the R-BUS drivers nor
with SawPro or SawStudio.

I wouldn't loose a lot of sleep over this.

There isn't a damn bit of difference between using the
three file formats, other than the packed file taking a
few extra cycles to massage.

Agreed. Note that the FAQ doesn't seem to say this either, which does not
make it wrong.

It'd be nice if I could take your word for that, but I
trust the CardDeluxe engineers a bit more than I trust
you.

Chevdo, that's a mistake that you get to make, and suffer for making it. I
guess you are unaware of the fact that books and FAQs are written by people
who may or may not be as fallible as people you find posting on Usenet.

But I do appreciate the response, because your
explaination of what the packed words were all about
sounds right and was informative to me. I hope someone
else responds to add more informative comments to the
subject..

Actually, just about everything that needs to be said, has been said. For
most of us, there's bigger fish to fry.

In article xlucg.26673$Fl1.12449@edtnps89, Chevdo wrote:

I can't really figure out what your fixation on file format is all about.

Because you keep talking about file formats. In fact, the file format
doesn't matter. Only the data format to the card does.
--scott


--
"C'est un Nagra. C'est suisse, et tres, tres precis."

Don Pearce wrote:
On Tue, 23 May 2006 21:13:15 +1200, "amp_noob"
wrote:

Rather common otherwise... Like confusing "your" and "you're"...

And "it's" or "its"...


You've all missed the worst one - "alot" for "a lot".

d



Well, there's always "They're".

-Raf


--
Misifus-
Rafael Seibert

blog: http://rafsrincon.blogspot.com/
Photos: http://www.flickr.com/photos/rafiii
home: http://www.rafandsioux.com