16 Bit vs 24 Bit

[junya-eskimo]

Hey ppl.

A friend of mine LIVES by the rule of bouncing/exporting all of his files as 24 bit..But has no real explanation to why its actually better..Just that it is (he, i wish you could meet him and watch him trying to explain)

I never though much into it..But he's hook on the idea..

My point was being that if i or the human race cant hear the difference then why does it exists..

The only thing i can think is that it plays a good role at a specific point of making a track somewhere from the start until going on cd..


Anyone out there wanna clear this up for me so i can 1 benefit from it and 2 stick it in my friends face later tonight...

Thanx all

Jun

[Tony Shepperd]

When I mix, I keep all the files at the original bit depth.
So if the session is 24/48 then that's what I deliver to mastering.
Try and keep the resolution as consistent as possible.

[lucey]

There is a sonic difference between 16 and 24 bits, in addition the math difference which helps further processing, so whatever you start with, use it all the way out.

In mastering, a digital guy will want your initial sample/bit rate. I dont really care what's sent but it's still best to stick with the all the bits you have.

Adding data, as a numbers game, that's a mistake.

[junya-eskimo]

Right ok, nice one...

Noted.Great to know that acctually......

Are there any Bit buster boys out there who can tell the benefits of using one or the other?

I am sure if 24 bit is an option it it must do something other than take up more hard drive space...ehee Or maybe not

cheerz

Jun

[lucey]

You're mostly looking at converter fidelity, and then plug in math, as the factors in sound quality that have nothing to do with skill.

The processing that happens ITB is usually over 24 bits, like 32, 48 or even 64. This numbers game is helped by a longer word at the outset.

24/44.1 is fine in my view. The projects I get that are 24/48k or 24/96k are no better unless the mixes and tracking were better. A weak mix is a weak mix and more time should be spent on skills, not numbers. Internal 48 or 96k has to be converted for rough mix CD burning, which is a lossy process for most people's software.

As for converters, The converter itself makes the bigger splash than the numbers you grab with it. A shitty converter at 24/96 will sound worse than 24/44.1 on a nicer one. So get a better converter, keep the sample rate low at 24 bits and off you go.


P.S. For future reference, this is not a 'High End' question thumbsup

[MeatPye]

A major advantage of 24bit is the greatly increased dynamic range. Very useful.

[lucey]

Originally Posted by MeatPye

A major advantage of 24bit is the greatly increased dynamic range. Very useful.

Well, maybe useful for 5% of the recorded music. Dynamics are more about the choices we make as mixers and at tracking. The main practicality is in processing down the line, and also yet a little less so, the initial fidelity at the converter.

[zulusound]

I mix OTB, using a Crane Song HEDD as A/D converter. Are you mastering guys saying you'd rather receive 44/16 mixes, as long as the converter (and mixes) are high quality? I usually deliver 96/24.

Evan

[lucey]

Originally Posted by zulusound

Are you mastering guys saying you'd rather receive 44/16 mixes,

Huh? No one said that.

All I'd say is turn off the processing, it gets in the way.


Sample rates have a different sound, but it's a bit of a Red Herring for 'quality'.

[zulusound]

You said this
"24/44.1 is fine in my view. The projects I get that are 24/48k or 24/96k are no better unless the mixes and tracking were better."

So, would you prefer 96/24 or 44.1/24 from the same good converter?

Evan

Originally Posted by lucey

Huh? No one said that.

All I'd say is turn off the processing, it gets in the way IMO.

[lucey]

Originally Posted by zulusound

You said this
"24/44.1 is fine in my view. The projects I get that are 24/48k or 24/96k are no better unless the mixes and tracking were better."

So, would you prefer 96/24 or 44.1/24 from the same good converter?

Evan

If it's ITB, then the depth/rate used ITB.

If there is analog processing on the 2 mix then I'm fine with 24/44.1. It's easier for you to make CDs and it's not a loss to choose 44.1 over 96k ... it's different. That difference is changed by any single move you make in the mix. The mix is most important, and the converter. The numbers are a distraction.

Each converter has a signature that is stronger than the sample rate. The sample rate is subtle, and the mixer compensates for that.

If we're talking tracking, I'd say 24/44.1 is fine. Many do 24/48k and I dont see an advantage worth the SRC that needs to happen for their rough CDs. 24/88.2 makes more sense as a balance of all qualities, but even then it's not as important as so many other factors.

[SpiderM69]

For sample rate education, read anything by Dan Lavry. His writings will give you a good dose of reality. Like anything else in life, there's tradeoff's with each choice.

[SpiderM69]

To address the original question, higher bit rate isn't just about greater dynamic range. To my understanding, more importantly it's about more bits giving better sample accuracy. Any EE's hanging around here can support or deny this, but it makes perfect sense to me.

[norman_nomad]

Originally Posted by SpiderM69

To address the original question, higher bit rate isn't just about greater dynamic range. To my understanding, more importantly it's about more bits giving better sample accuracy. Any EE's hanging around here can support or deny this, but it makes perfect sense to me.

1 bit can capture, with perfect accuracy about 6db of dynamic range. Adding more bits does not increase fidelity. 24 bits allows for 144 db of dynamic range which is larger than any converter can physically (i.e. in the real world) resolve, so there is no need for a larger bit depth.

Larger bit depths are used ITB to allow more drastic gain staging and processing to be executed without degradation, but not because 32bit or 48 bits sounds "better".

[SpiderM69]

I should have been more specific in my response. I was referring to 16 bit vs 24 bit.

[norman_nomad]

Originally Posted by junya-eskimo

Hey ppl.

A friend of mine LIVES by the rule of bouncing/exporting all of his files as 24 bit..But has no real explanation to why its actually better..Just that it is (he, i wish you could meet him and watch him trying to explain)

I never though much into it..But he's hook on the idea..

My point was being that if i or the human race cant hear the difference then why does it exists..

The only thing i can think is that it plays a good role at a specific point of making a track somewhere from the start until going on cd..


Anyone out there wanna clear this up for me so i can 1 benefit from it and 2 stick it in my friends face later tonight...

Thanx all

Jun

Exporting files that are to be processed or used later in a mix at the "destination" bit depth is generally a good practice, but as lucey points out, this is probably one of the lesser things to worry about in the grand scheme of a mix.

Bit depth determines the maximum dynamic range of the source file. 16 bits allows for 96db of dynamic range. This only affects the music as it fades into quantization noise floor and if the file has been mixed down to 16bit and not dithered properly, you may here distortion and other nasty bits. (Although even if undithered much of this is often masked by gear in the signal chain with self-noise far above the -96dbfs threshold).

Best practice if your making stems is to output to the destination bit depth and use a non-shaped dither when exporting both 16bit and 24bit files.

[Bob Olhsson]

It's a good idea to save anything that's going to be processed or edited further as a 24 bit file. With today's drives costing well under a dollar a gigabyte, there's no sensible reason at all to save something as 16 bit.

[junya-eskimo]

Right cool...Its kind of what i thought...Its good to know about the db range between the 16 and 24 bit....That will come in handy...

Thanx alot for you input on this..

Just as a last little question...I will be running from a mac Logic pro to an Apogee DA-16X to a NEVE Summing mixer then back into a RME 800 ...Is there any thing i should know about? ANy tips or no no's ?

If not all good and thanx for all the info

cheerz

Jun thumbsup

[danger]

Yo Junya,

My 2 cents, when you have an Apogee and a Neve already its worth to replace the rme to a crane song hedd or a lavry A/D

[EliasGwinn]

Not to beat a dead horse...

The bit-depth gives you dynamic range AND signal-to-noise ratio (SNR). The difference is the SNR describes where the noise floor is, and dynamic range describes the lowest possible tone that is resolvable (resolution), which is usually well below the noise floor.

24-bit is important for these low-level resolutions, EVEN WHEN THE MUSIC HAS LITTLE DYNAMICS. Whether or not an iPod will resolve the tail-end of a reverb or not doesn't mean you don't need to preserve these details for the sake of the recording.

And its always better to keep the SNR as maxed out as possible, even if you do like the sound of analog tape. If you send a 16-bit track to the mastering house, you've already (hopefully) added dither, and they will have to as well. All this noise really adds up, and it is very audible.

Bob Katz talks (very clearly and intelligently) about this on his site:

Digital Domain - Debate about Dynamic Range, Bit Depth, ...

Enjoy!!

[AllAboutTone]

I have both on different machines meaning different converters as well. YES the noise floor is less with 24 bit, thats not a issue at all, after using both for a while now i have mixed feelings, i do all my rock and roll with 16 bit, love that punch, however when it comes to voice over or detail quiet tracks i reach for the 24 bit.
I use 16 bit much more overall. FWIW.

[imaverysoundman]

didnt read all the thread so maybe someone answered this.....
(got this info from bob katz mastering book)............

multiply the bitrate by 6 (dont know why its 6....but it is!!!) to give the maximum possible dynamic range...
therefore 16(bit)x6=96...therefore if you record at 16bit then the maximum dynamic range you can possibly have is 96db.............
however the human ear is capable of a 120db dynamic range...
enter 24bit.....
24x6=144....therefore if you record at 24bit you go 24 db below the human hearing range.........point being.....you dont have to record all your signal levels hot..ie.. 0db (as im sure alot of inexperienced engineers do.....including me untill i understood this!!!!!)........
then when you master and push it up an extra 10db(or wharever it works out to be) you dont introduce any extra noise to the signal..thus keeping the track as clean as possible............
think of it as bitrate =y axis, samplerate =x axis
sample rate helps keep the signal as smooth(analog) as possible..(helps reverb tails and fades etc)
bitrate helps reduce noise and help dynamics...

which dosnt really matter a **** anyway coz it all gets mastered to rediculous levels taking away any dynamic that may have been captured...........

hope i explained that ok.....it was a revelation to me when i finally understood it....

rock on!!!!!!!!!!

tris..................................

[AllAboutTone]

Originally Posted by imaverysoundman

didnt read all the thread so maybe someone answered this.....
(got this info from bob katz mastering book)............

multiply the bitrate by 6 (dont know why its 6....but it is!!!) to give the maximum possible dynamic range...
therefore 16(bit)x6=96...therefore if you record at 16bit then the maximum dynamic range you can possibly have is 96db.............
however the human ear is capable of a 120db dynamic range...
enter 24bit.....
24x6=144....therefore if you record at 24bit you go 24 db below the human hearing range.........point being.....you dont have to record all your signal levels hot..ie.. 0db (as im sure alot of inexperienced engineers do.....including me untill i understood this!!!!!)........
then when you master and push it up an extra 10db(or wharever it works out to be) you dont introduce any extra noise to the signal..thus keeping the track as clean as possible............
think of it as bitrate =y axis, samplerate =x axis
sample rate helps keep the signal as smooth(analog) as possible..(helps reverb tails and fades etc)
bitrate helps reduce noise and help dynamics...

which dosnt really matter a **** anyway coz it all gets mastered to rediculous levels taking away any dynamic that may have been captured...........

hope i explained that ok.....it was a revelation to me when i finally understood it....

rock on!!!!!!!!!!

tris..................................

Very good post, never knew that, thanks.
However most of my bands/clients never have there project mastered so i give them as much dynamics with my 2 buss poor man mastering as i can before breaking up.

[RecTeach]

I haven't been on here in awhile... is this how far the "High End" section has fallen??? Where we are discussing why 24bit is better than 16? Please tell me things haven't gotten this bad...

[jayfrigo]

Originally Posted by SpiderM69

To address the original question, higher bit rate isn't just about greater dynamic range. To my understanding, more importantly it's about more bits giving better sample accuracy. Any EE's hanging around here can support or deny this, but it makes perfect sense to me.

For 16 bit vs 24 bit, the top 96 dB of the range isn't actually any better with the 24 bit system. What you gain is an extra 48 dB below the 96 dB floor of the 16 bit system. That is the additioanl detail; it is low level information below the LSB of the 16 bit system.

Interestingly, the last few bits of the 24 bit system actually are not useful in an analog conversion sense since we have no analog electronics capable of more than around 21 bits equivalent performance. This is not to say that 24+ bits aren't useful and important in processing, just that analog electronics are limited to a best case scenario of about 21 at the conversion stages (though even this is seldom achieved).

It is also worth remembering that the the extra dynamic range is not additional headroom, rather additional "footroom." 0 dB FS is still the top of the range for 24, or 16, or 8 bits for that matter. 0 dB FS is zero decibels full scale, or all bits on, the maximum signal that can be represented. The extra dynamic range comes at the bottom of the range, below -96 dB in the 16 bit system, or below -48 dB in the 8 bit system.

And with the 8 bit system, the top 48 dB is represented just as well as on the 16 or 24 bit system. Of course it sounds like crap by comparison since the noise floor and quantization distortion is now only 48 dB below full scale, which is plainly audible to the detriment of any complex musical signal. However, it in no way reduces the properly implemented 8-bit system's ability to perfectly capture the top 48 dB, deficient though that range may be for quality music recording.

[Brad McGowan]

Thanks for clarifying Jay! That makes total sense and explains why 16-bit recorded tracks sound flat and 2D to my ears.

Brad

[joelpatterson]

Originally Posted by RecTeach

I haven't been on here in awhile... is this how far the "High End" section has fallen??? Where we are discussing why 24bit is better than 16? Please tell me things haven't gotten this bad...

It's probably even much worse than you think.

[midnightsun]

A point to consider here is that most converters today convert at 24 bit. A file that is 16 bit is probably a file that has been dithered down to 16 bit. Not all dithering is created equal.

I agree with what danger said about pumping your final analog signal into a Cranesong or Lavry box. I also come out of my Logic DAW with 16DAs to an analog console. This my final analog stereo mix goes to a Hedd 192. With the Hedd I can print my final stereo mix at any 16 or 24 bit and from 44.1KHZ to 192KHZ. For the most part I am making demos that need to be played on Redbook CDs; thus, I need 44.1/16. The Hedd allows me to monitor the 44.1/16bit sound as I am printing. It seems like I am mixing to what it is going to sound like at 16 bit. I can hear a subtle difference between the 16 and 24 bit on the Hedd. What blows my mind about the Hedd is that printing at 44.1/16 sounds better to my ear than other converters at higher SR and 24 bit. In short, I am very satisfied with 44.1/16 coming out of the Cranesong and use it 95% of the time.

[AllAboutTone]

wow, there is a lot of info to learn here, again i like the best of both worlds, 24 bit to me is more balanced but almost has a compression feel to it somewhat, i cannot explain it and punch, i can only hear it with my own ears. 16 bit just seems to have more iron in it, i know its noise by all means.
im not sure that it matters a whole lot since it will end up 16 bit on the final master or sould i say on a myspace page...lol

[norman_nomad]

Originally Posted by imaverysoundman

.....you dont have to record all your signal levels hot..ie.. 0db (as im sure alot of inexperienced engineers do.....including me untill i understood this!!!!!)........
then when you master and push it up an extra 10db(or wharever it works out to be)

Actually, what Bob is saying is that when the signal to noise ratio is poor and then you raise the gain levels later, you push the noise floor up as well. So he's actually advocated that you record hot (at least on a 16bit system... kind of outdated info) and reasonably hot on a 24bit system to get the signal as far away from the noise source (alias-type distortion) as possible.

Originally Posted by Bob Katz (from link posted above)

I would still
recommend (as a perfectionist) peaking the original A/D conversion near to 0 dBFS.
Otherwise you are just adding the non-linear low level alias-type distortion of the
crappy A/D to the already distorted original crappy source :-)

What Bob maybe doesn't emphasis enough is how unimportant it may be to max record near absolute zero on a 24bit system. In real world practice the quantization noise floor of a good 24bit converter is low on the list of noise issues to worry about relative to the noise of your preamp, mic and ambient noise of the recording environment.

In real life, I know I've had more problems trying to salvage some guy's clipped tracks than I have trying to salvage tracks where the quantization noise floor of the converters crept too far into the source signal.

So please, on a 24bit system, don't record so hot that you're likely to clip!