bit depth/ sample rate question ???

[BFSound]

Hi,

When doing a location recording or any recording in general, what are the standards for the following?

recording, then mixing, mastering or final end product.

Is it best to record 24/48 for mixing and then mixdown to 16/44? then leave it at that for the end product?

i understand bit depth and sample rates and they're effects but it seems you would want 24bit for mixing and then 16 for the end result?

Thanks

[Audiop]

Typically you record and mix in high resolution and when you have nailed the sound you step down to the end format for example redbook CD.

Going down from 24/96 to 16/44.1 (redbook) you first convert sample rate and then you add dither and then truncate the wordlength of the 24 bit samples to 16 bit samples.

If you leave the mix to a mastering guy you should give him the 24/96 material and let him do the SRC and wordlength reduction.


/Peter

[boojum]

OTOH it was recently discussed in another thread that a group of engineers and artists could not distinguish 44.1 from higher bit rates in a double blind test so the studio just went ahead and recorded 44.1.

If there are double blind tests where the tested could tell the recordings apart I would like to know where the tests are. It would end this discussion forever.

[Audiop]

Some converters are cleaner at higher rates even if you can not hear the extended bandwith as such.



/Peter

[joelpatterson]

Originally Posted by boojum

... if there are double blind tests where the tested could tell the recordings apart ....

Ah, but then there's our old nemesis, testing bias... what about the argument someone could put forward (me, if the price was right) that 48 just feels, psychologically, like it ought to be better, and that's enough of a reason. The positive placebo reinforcement theory.

[boojum]

Originally Posted by joelpatterson

Ah, but then there's our old nemesis, testing bias... what about the argument someone could put forward (me, if the price was right) that 48 just feels, psychologically, like it ought to be better, and that's enough of a reason. The positive placebo reinforcement theory.

Joel, that is what te double-blind test is designed to eliminate. Neither the testor nor the testee knows which sample is which.

[chris319]

The "blind" in double blind means the guinea pigs, er, test subjects, don't know what sample rate they're hearing at any given time, or in medicine whether they're receiving the real drug or placebo.

I once did a triple-blind test involving three mics. I posted recordings of an SM58, an SM57, and a cheap brand "B" dynamic mic (not Beyer and not B&K). The idea was to see if people could identify the cheap mic. People started demanding to know what kind of acoustical treatment there was in the room and what kind of preamp I was using. I said all they needed to know was that they were recorded in the same acoustical environment with the same preamp in the same session. Well then they got all mad and one guy who made himself out to be some kind of grand high exalted expert was convinced I was using a speech synthesizer (I used a live male announcer). The on-line "experts" got their panties all in a bunch and none of them would so much as hazard a guess. Finally they got so huffy and puffy that the moderator had to close the thread

[boojum]

Originally Posted by chris319

The "blind" in double blind means the guinea pigs, er, test subjects, don't know what sample rate they're hearing at any given time, or in medicine whether they're receiving the real drug or placebo.

I once did a triple-blind test involving three mics. I posted recordings of an SM58, an SM57, and a cheap brand "B" dynamic mic (not Beyer and not B&K). The idea was to see if people could identify the cheap mic. People started demanding to know what kind of acoustical treatment there was in the room and what kind of preamp I was using. I said all they needed to know was that they were recorded in the same acoustical environment with the same preamp in the same session. Well then they got all mad and one guy who made himself out to be some kind of grand high exalted expert was convinced I was using a speech synthesizer (I used a live male announcer). The on-line "experts" got their panties all in a bunch and none of them would so much as hazard a guess. Finally they got so huffy and puffy that the moderator had to close the thread

Chris, if the person administering the test knows it is a single blind test. If neither the person administering the test nor the person taking the test know it is a double blind test. There are sometimes unconscious subtle cues given if the person administering the test knows what the item is he is offering the person taking the test.

A fellow I know on another board administered a test to a fellow, a self-proclaimed "Golden Ears," who swore he could easily identify MP3's from WAV files. The fellow was given a dozen files to determine the type from. He quite proudly and surely listed which were WAV and which were MP3. They were all MP3, 128KBit, too. The fellow who was tested never spoke to the fellow administering the test again.

This is only one case but it is so easy to fall into the ego trap of "knowing." There is also a reason that clinical testing uses double blind testing. It rules out bias to a great degree if not totally. I find double blind tests way easier to trust. In the world of audio they are regarded with more and more suspicion as the price of the gear goes up. Go figure. Just my not-so-humble opinion.

L8R

[joelpatterson]

Originally Posted by boojum

Joel, that is what te double-blind test is designed to eliminate. Neither the testor nor the testee knows which sample is which.

But, see, I am questioning the whole premise of ANY test: the idea that you put someone on the spot, ask them to distinguish some "betterness" between two peculiar, specific things-- and then you draw overarching conclusions for all time and across continents.

Sure, this is done all the time and is the basis for much that goes on in life in general... I just wonder if it's the best approach, that's all....

[Rainy Days]

Originally Posted by Audiop

Typically you record and mix in high resolution and when you have nailed the sound you step down to the end format for example redbook CD.

Going down from 24/96 to 16/44.1 (redbook) you first convert sample rate and then you add dither and then truncate the wordlength of the 24 bit samples to 16 bit samples.

If you leave the mix to a mastering guy you should give him the 24/96 material and let him do the SRC and wordlength reduction.


/Peter

Great answer.

[daveg62]

I always record 24/44.1 ,I feel it's more important to have decent sounding converters not what rate you record at.
Here's a great read...I tend to think this guy knows a little more on the subject than most.
http://www.lavryengineering.com/docu...ing_Theory.pdf

I like this quote the best
"The optimal sample rate should be largely based on the required signal bandwidth. Audio
industry salesman have been promoting faster than optimal rates. The promotion of such ideas
is based on the fallacy that faster rates yield more accuracy and/or more detail. Weather
motivated by profit or ignorance, the promoters, leading the industry in the wrong direction, are
stating the opposite of what is true."

[BFSound]

Originally Posted by daveg62

I always record 24/44.1 ,I feel it's more important to have decent sounding converters not what rate you record at.
Here's a great read...I tend to think this guy knows a little more on the subject than most.
http://www.lavryengineering.com/docu...ing_Theory.pdf

I like this quote the best
"The optimal sample rate should be largely based on the required signal bandwidth. Audio
industry salesman have been promoting faster than optimal rates. The promotion of such ideas
is based on the fallacy that faster rates yield more accuracy and/or more detail. Weather
motivated by profit or ignorance, the promoters, leading the industry in the wrong direction, are
stating the opposite of what is true."

Thanks, i will be using the studiolive 24.4.2 which records 24/ 44 or 48, after i get it where i like it i should just leave it there if i want someone to master? otherwise i would just mixdown to 16/44 for CD to listen to.

Thanks

[rumleymusic]

Thanks, i will be using the studiolive 24.4.2 which records 24/ 44 or 48, after i get it where i like it i should just leave it there if i want someone to master? otherwise i would just mixdown to 16/44 for CD to listen to.

In this case bit depth is much more important than sample rate. 48kHz allows you to record a fraction of an octave higher than 44.1, but 24 bit is 256 times the resolution of 16 bit. Leave it the way it is for the mastering at 24/44.1 or 48, and never reduce the bit depth to 16 bit until the product is finished. If you want to burn yourself a reference CD, that is one thing, but don't give that to an ME.

[JTraverse]

Keep the sample rate at 44.1kHz. It's a general rule of thumb to use a sampling rate which is around double to that of the highest frequency we hear, which is approximately 20 - 22 kHz. Unless you are making animal music, don't get caught up in worrying about it.

192kHz and things are just quite useless (unless we are talking plug-in oversampling, which is totally different), and the higher the sample rate, the more you induce aliasing. Google that one.

Your bit depth is what matters. 24 or even 32 floating point bit resolution is great and then a high quality dither can used to bring it down to 16 bit at the mastering stage.

[BFSound]

Originally Posted by rumleymusic

In this case bit depth is much more important than sample rate. 48kHz allows you to record a fraction of an octave higher than 44.1, but 24 bit is 256 times the resolution of 16 bit. Leave it the way it is for the mastering at 24/44.1 or 48, and never reduce the bit depth to 16 bit until the product is finished. If you want to burn yourself a reference CD, that is one thing, but don't give that to an ME.

thumbsup thanks

[BFSound]

Originally Posted by JTraverse

Keep the sample rate at 44.1kHz. It's a general rule of thumb to use a sampling rate which is around double to that of the highest frequency we hear, which is approximately 20 - 22 kHz. Unless you are making animal music, don't get caught up in worrying about it.

192kHz and things are just quite useless (unless we are talking plug-in oversampling, which is totally different), and the higher the sample rate, the more you induce aliasing. Google that one.

Your bit depth is what matters. 24 or even 32 floating point bit resolution is great and then a high quality dither can used to bring it down to 16 bit at the mastering stage.

Sounds good i think you two answered my question perfect!

[mixedupsteve]

Originally Posted by JTraverse

hen a high quality dither can used to bring it down to 16 bit at the mastering stage.

This is what Izotope reccomends. Dither at the future bit depth of the final product.

[Ozpeter]

The previous discussion at Bit depth revisited left a fair bit of blood on the floor.

I note the OP asked about standards, which perhaps differ from individual preferences and personal opinions. So this time round I shall say nothing!

[AlexK]

Originally Posted by Audiop

Typically you record and mix in high resolution and when you have nailed the sound you step down to the end format for example redbook CD.

Going down from 24/96 to 16/44.1 (redbook) you first convert sample rate and then you add dither and then truncate the wordlength of the 24 bit samples to 16 bit samples.

If you leave the mix to a mastering guy you should give him the 24/96 material and let him do the SRC and wordlength reduction.


/Peter

Recording at 96kHz S/R and then converting the sample rate to 44.1 (which ISN'T a factor of 96) is actually more destructive than recording straight to 44.1. It's fairly basic maths.

If you really MUST record at higher sample rates, stick to 88.2 (as 44.1 is exactly half, so two samples are converted to one sample, as opposed to 2.1768 samples being converted to one).

If you double the sample rate, then you double the bandwidth and CPU load on the computer...

[ghellquist]

Originally Posted by AlexK

Recording at 96kHz S/R and then converting the sample rate to 44.1 (which ISN'T a factor of 96) is actually more destructive than recording straight to 44.1. It's fairly basic maths.

Sorry, but you are totally wrong there. Most often at least in todays world.

A good algorithm today, written for todays very capable computers, will not have problems with any ratio between the two sample rates. Older algorithms often were restricted by the available raw CPU power and had to take shortcuts. Not so today if you know your material. You can of course find older algorithms in some of todays modern programs. And some programmers are not very good at it.

The traditional solution in mastering is to DA convert, treat with analog gear and finally AD-convert. In that chain there is no relation between the two sample rates.

// Gunnar

[rumleymusic]

Recording at 96kHz S/R and then converting the sample rate to 44.1 (which ISN'T a factor of 96) is actually more destructive than recording straight to 44.1. It's fairly basic maths.

There is kind of a myth that if you want a 44.1 destination you should record at 88.2 and the computer will simply have to divide by 2. This is not how modern SRC's work, or how math works in a computer. So it does not matter if you record at 48, 88.2, or 96. Your judgment on which sample rate you choose should be determined by the performance of the converters at that sample rate, and the positive effects of the added harmonic overtones you capture.

Just don't use a crummy SRC.