Is 16BIT enough vs 24BIT

filmelife

filmelife

Well-known member
According to audio books 24bit gives you a higher dynamic range and more bits to apply processes to work with. Then you have audio pros that say 16bit is plenty,
i dont understand how 16bit is plenty after applying noise reduction, compression, equalization etc.

can we put this to rest, is 16bit enough? Would i be correct by saying 16bit is not enough for a theatrical?
​
 
24 bits is good to start with, especially if you need to raise levels in post or do a lot of manipulation.

But again, clean 16 bits gives you more dynamic range than most (all) systems can deliver in real life. And cheap 24 bit recorders do not give nowhere real 24 bit dynamic range anyway, even the very best like SD7xx series are only about 19-20 bits in real life (about 110-115 dB S/N ratio). In thise sense 24 bit is overvalued, even if it says 24 in the file header the signal can be just about anything in quality, specially with cheaper systems.

So there is no clear truth, carefully recorded 16 bits is better than reproduction systems can deliver and most people hear, but it never hurts to have the extra handling space of 24 bit originals.

All better editing systems do the calculations at 32 bits also for 16 bit signals, so in that sense there is no advance in using 24 bits.

More than 16/24 the quality of the originals matter; high levels without clipping, low noise preamps, nice limiters.
---------

Is 16 bits enough? From another angle: all pre-digital audio in the great movies of the fiftes, sixties, seveties, eighties and even early nineties is only "12-14 bits". Was that enough? I think they still sound great. Now we are afraid or unable to use vastly superior 16 bit digital audio because "it is not good enough"...

But like I said, use 24 bits if you really have it, does not hurt any, just one down conversion at the end.
 
Last edited:
wow,nice reply. I just wonder if its worth the extra effort to have a sound recordist recording at 24bit, when i could have him doing another duty like assisting.
I could just run the 302 straight into the HVX and get clean dialogue. Post editing in 32bit helps a lot, professionals claim that they dont use anything less than 24bit for theatrical productions?
 
If you've got a dedicated sound guy, use him for that. Don't neglect audio; it's the one thing which will kill your project quicker than anything else if it's bad.
 
I'd first like to say again that I'm not a production recordist so those that are more experienced in this please contradict me. I don't see how recording at 24 bit would take any more effort than it does to record at 16. The only drawback with 24 bit is that it results in larger file sizes, but hard drive space is so cheap these days that's practically a non issue. If you are thinking more along the lines that you want to not have a Sound Mixer and just run the audio into the camera, then I would suspect that sound quality is not very high on your list of priorities. I would think that it would be hard for the camera operator to pay strict attention to sound as well as the image ... I can't even type and time same at gum the chew.
 
what more important is your analog stage, so better preamp with 16 bit recorder would sounds better then poor preamp with 24 one
- even i have 24 bit recorder i never use it for dialog as it completely waist of space ,
for music use 24 for better dinamics and resolution .
 
There are a lot of "professionals" out there and some of us are just as full of $%^& as everybody else.
I don't know what the standard is right now for post but I highly suspect that everyone has moved over to 48/24. I know that a number of FX recordists record 96/24 almost all the time. My understanding is that production is recording most often at 48/24.

There are a number of reasons to record at 24 bit. True since your not maxing out you are recording at maybe 20-22 (effective bits) if your doing a really good job. But the same is true for 16 bit so then your recording at an effective bit rate of 12-14. So if you want 16 bit resolution record at 24. Another reason for 24 bit and why most people I know are using it in post is that when you apply a process to the sound you have more to work with and it is smoother, sounds better.

In your case you want to go direct to camera. Depending on the camera that could be dropping you way down in quality. Many cameras have crappy sound sections.

Dolby digital is 44.1/16 (and compressed) so in terms of final out put you going to a lower spec than any of the formats were talking about.
We did a bunch of listening tests with my Marantz (Mix Pre front end) and a Nagra V. We could hear the difference between 16 and 24 easily. 48 VS 96 was just noticable on the right material with the Nagra and pretty much not noticeable at all with the Marantz. 44.1 and 48 I have never heard any difference with. Film folks (SFX) mostly recorded in 44.1 because that was the end format and conversion used to take a long time. Now most are recording at 48 because of 48 being the DV format. There is a slight argument for 96 in FX work because you are often slowing things down and 96 gets a lot less "gritty" but I think a lot of it is "my recording is bigger than yours", something that always seems to be a factor in this biz. There are a lot of folks who will argue all day that X is the best and anything else is ****. Generally ignore those folks. There are a lot of ways to get great sound, most of it has to do with the person operating the box not the box itself.
 
manipulating with sfx is always better with higher sample rate - for dialog i bet you will never know inside complete mix what rate it was recorded with , self noise floor and set noise are more then enough to cover the low end of the resolution :-) , waist of space
by the way i can guarantee you the mp recording in 320 for dialog would be good enough as well if you have good analog and ato d converter as the jvc have in their cameras - dont sure you will hear the compression at least i people who heard it could say it is not good for cinema release (checked side by side with sd 744in 16 bit)
 
When I said 24 is actually 19-20 bits it is the mic and mic preamp noise which limits the S/N ratio to about 110 dB (about 18 bit dynamic range) even with the best systems, not the fact that peaks do not hit 0 dBFS levels. With the same quality preamps we get clean 96 dB S/N ratio with 16 bits, thus the difference between 16 and 24 bit systems is only about 12 dB at best, which corresponds to 2 bits more real dynamic range. Even that is questionable, as most mics have a lot less dynamic range. So, most of the time, having 24 bit files only makes the recordest feel good, not the sound any better. But it does not harm anybody either.
 
very interesting. I gather from all the comments that

-if a few processors must be applied to dialogue, 24bit would be better

-if recording for theatrical features, 24bit would also be better for that extra dynamic range that a theatrical feature can afford

-if its a basic recording without any processors needed (unlikely) then 16bit should do

-in either case, a seperate recorder should be used to maximize quality, better preamps than the camera, more bits to work with just in case

-if a production is picked up by a distributer, then the audio mix taken to a facility will be better at 24bit for the mixers/editors to work with.
 
killfilm said:
I could just run the 302 straight into the HVX and get clean dialogue.
Click to expand...



I've done a significant amount of testing using a MixPre feeding both HVX/DVX and a MicroTrack at the same time... and the difference is a lot bigger than I'd expected... the MicroTrack sounds a LOT better - the DVX/HVX generates a lot of electrical noise that really stands out (in comparison) in the quiet parts. This of course is NOT a sample size issue... just an issue of the analog stages being more important...
 
Petrus,
I wont argue the math. There were a lot of folks who swore they could tell the difference between 48 & 44.1, I never could. In blind tests we could all hear the difference between 16 and 24 bit recordings. Now that may be because your always going to be recording with headroom and with 24 bit were actually recording what an optimized 16 bit recording would be? But it was very apparent. I also record chamber concerts and when I switched to 24 bit recordings everyone commented on it and none of them knew I had changed anything. So math aside I always record 24 bit. But YMMV. As to dialog. I said I thought that 24 was becoming the standard. But that in no way implies it ought to be the standard. I don't know if 24 is better for dialog, I sort of suspect that it doesn't make much difference. 24 is better for post since they will be doing processing and it's an advantage there but that doesn't mean you have to record at 24.
Another thing that should be mentioned is that 17 bits is not "one more" than 16 bits it's twice the resolution. Every bit doubles your resolution so in theory an optimized 24 bit recording is 256 TIMES as much resolution as an optimized 16 bit resolution. Even if your S/N is still only 96dB (because of preamp and mic self noise) , it's a smoother 96dB.

Killfilm,
2- not really. Theatrical features don't nec have any more dynamic range than a DVD.

3- 16 bit should do nicely either way. Post can work in 24 bit even if the recordings were at 16.

4- generally yes. But some cameras have very good audio sections.

5- not nec. see 3 above.

Part of the reason I think production is leaning toward 24 bit is because more and more productions are on some kind of HD/Flash recorder. Even if it's a waste of space it doesn't really matter, there is plenty of space. Post is converging on 24 bit because more processing is done digitally and much less is being done analog and in the digital realm there is an advantage to 24 bit. FX are going to be at 24 bit (probably) and music is almost certain to be at 24. It makes things smoother when everybody is in the same format so the pressure is going to be to make everyone work at 24 whether it makes a difference or not. This is what kept a lot of places at 44.1 for so long.
But in the long run it's what it sounds like. Black Stallion was noted for some great water sounds. Many of those were recorded on a Sony portable cassette deck with the mic that came with it. Good source, good recordist, and used artfully, all way more important than the tech spec.
 
Scott, very true about your analysis.

Also, consider that 24-bit helps a lot in the mixing stage. It's okay to mix 10 tracks and come out clean, but once you get into say 32/48/96 tracks the amount of noise present will be very noticeable. The Signal-to-Noise ratio from using 24-bits is going to make a huge difference once you downmix every element to 2/6/12 channel.
 
Noiz2 said:
Another thing that should be mentioned is that 17 bits is not "one more" than 16 bits it's twice the resolution. Every bit doubles your resolution so in theory an optimized 24 bit recording is 256 TIMES as much resolution as an optimized 16 bit resolution. Even if your S/N is still only 96dB (because of preamp and mic self noise) , it's a smoother 96dB.
Click to expand...

Not so. It is true that 17 bits means twice the dynamic range compared to 16 bits, 6 dB more that is. It is still dynamic range, not "resolution". If you have 96 dB of true dynamic range (almost impossbile to get in real life situations) it can be fitted to 16 bit system, and having 24 bits with same peaks would just add 48 dB worth of highly detailed hiss at the bottom, the 96 dB at the top is IDENTICAL in both systems.

My point is that 16 bits properly used is so good that it practically always surpasses the source, reproduction and precieving qualities of the audio chain from location to brains (noise floor at location, mics, analog electronics, loudspeakers, viewing room, human hearing). Using 24 bits, but only with great analog front end, gives sound engineers easier times only. For that reason it in not something to loose sleep over.
 
Petrus - sorry, you're still confused about this. RESOLUTION is the major advantage higher sample size. Every bit adds a logarithmically-smaller extension to the dynamic range. 24bit sampling gives you 256 times the resolution throughout the dynamic range, and a about 1/30,000 extension to the measurable dynamic range.

The entire dynamic range of any digital audio measured in dBFS is exactly the same... 0 to 1. A one-bit PCM system has a dynamic range of 0-1, as does 8 bit, 16-bit, or 24 bit. One is always the loudest representable sound. 0 is always no sound. More bits just add resolution to our measurement of signals falling between 0 and 1. Extra bits determine the smallest representable quantity, not the largest - the largest is always 1 (which we call 0dBFS).

Petrus, if you really believe that the additional 8 bits affect only the part of the ignal below -96dB, can you please explain the numbering system that would achieve this? 'Cause no encoding I've heard of (or can imagine) would have that property and still be useful for arithmetic operations.
 
Last edited:
Here's how the binary representation of digital audio works - on the left is the binary, on the right is the fraction of 0dBFS.

1-bit:
0
1

2-bit:
00 0
01 1/3
10 2/3
11 1

4-bit
0000 0
0001 1/15
0010 2/15
0011 3/15
0100 4/15
0101 5/15 = 1/3
0110 6/15
0111 7/15
1000 8/15
1001 9/15
1010 10/15 = 2/3
1011 11/15
1100 12/15
1101 13/15
1110 14/15
1111 1

16-bit:
0000000000000000 0
0000000000000001 1/65535 (-96dB)
0000000000000010 2/65535
....
0101010101010101 21845/65535 = 1/3
....
1010101010101010 43690/65535 = 2/3
....
1111111111111100 65532/65536
1111111111111101 65533/65535
1111111111111110 65534/65535
1111111111111111 1


24-bit:
000000000000000000000000 0
000000000000000000000001 1/16,777,215 (-144dB)
000000000000000000000010 2/16,777,215
...
000000000000000100000000 256/16,777,215 = 1/65535 (96dB)
...
010101010101010101010101 5,592,405/16,777,215 = 1/3
...
101010101010101010101010 11,184,810/16,777,215 = 2/3
...
111111111111111111111100 16,777,212/16,777,215
111111111111111111111101 16,777,213/16,777,215
111111111111111111111110 16,777,214/16,777,215
111111111111111111111111 1

As you can see, additional bits add resolution throughout the range. They don't extend the range.
 
Last edited:
It is the other way around, adding bits only extends the dynamic range.

We get 0 dBFS when all the available numbers are used up (1111 1111 1111 1111 on your 16 bit sample). We hit the sampling noise floor when there are no numbers to use anymore (last bit acts randomly). Each additional bit in the system doubles the number range in binary system (just like adding one more digit in normal decimal system makes the range 10 times bigger), which means that twice as big voltage can be presented. This is where the 6 dB additional dynamic range for each additional bit comes from, twice the voltage = 6 dB more DR.

I think we tried to discuss this once before to no avail. The above is the most simple explanation for a basically simple system I can come up with.
-----------

Some snippets from Wikipedia (article: Audio bit depth):

"By increasing the sampling bit depth, smaller fluctuations of the audio signal can be resolved (also referred to as an increase in dynamic range). The 'rule-of-thumb' relationship between bit depth and dynamic range is, for each 1-bit increase in bit depth, the dynamic range will increase by 6 dB. 24-bit digital audio has a theoretical maximum dynamic range of 144 dB, compared to 96 dB for 16-bit."

and

"The importance of bit depth in PCM audio is that it determines the maximum possible dynamic range of the signal. For a typical PCM recording, in which no noise shaping is employed and the frequency range extends most of the way to the Nyquist Limit, the dynamic range in dB is equal to 1.92 + 6.02 * bits. This formula is often simplified to 6 dB per bit, which yields the common value of 96dB for 16 bit CD audio."

Sound Devices has also something to say about this (what do THEY know?):

"Bit Depth = Dynamic Range.

Bit depth defines the digital “word length”used to represent a given sample. Bit depth correlates to the maximum dynamic range that can be represented by the digital signal. Larger bit depths theoretically yield more dynamic range. A quick estimate of maximum dynamic range capable of being represented by a given word length is dynamic range ~= no. of bits x 6 dB. Bit depth is an exponential measure (exponent of 2), so as bit depth increases, theamount of data it represents increases exponentially. The majority of field recording is done with 16-bit audio, therefore,each sample is represented by a digital word of 2^16 (65,536) possible values. 24-bit audio has a word length of 2^24 (16.7 million) possible values per sample."
 
Last edited:
Petrus said:
Each additional bit in the system doubles the number range in binary system (just like adding one more digit in normal decimal system makes the range 10 times bigger), which means that twice as big voltage can be presented.
Click to expand...

Petrus, this is where you are confused. Sorry! In digital audio formats, 0dBFS is always the maximum voltage. You never double the largest voltage that can be represented. When you add a bit to the representation, you HALVE the SMALLEST voltage that can be represented. That is why dBFS values are negative.

Here's an experiment for you. Take a preamp and a converter that runs at either 16 or 24 bits. Set the converter for 16bit operation. Calibrate the preamp with the converter so that when the preamp puts out a signal of, say, 20dBV, the converter registers 0dBFS. Turn up the gain on the preamp so that it ouputs, say, 26dBV. The converter registers a digital peak.

Now switch the converter to 24bit operation, leaving all else unchanged. Does it still peak when the preamp puts out 26dBV? Or, by your theory, will it accept s signal level of 68dBV now before registering a peak?

No. The converter will consider 20dBV to be 0dBFS (peak level) in either case. The top end of the scale is fixed.


Anyway, like I said, please explain a numbering system that will act the way you claim - adding range but not resolution in the dBFS scale...
 
I don't know where you SD quote comes from - but http://www.sounddevices.com/notes/recorders/real-world-24-bits/ clearly refers to "16 vs 24 bit RESOLUTION" and the advantages not in dynamic range, but in how increased resolution allows more forgiveness of using less of your dynamic range:

"Leaving this fact aside, let’s look at what 24 bit recording can do in real world environments with high performance gear… 24 bit resolution versus 16 bit resolution is a genuine advantage. 24 bit recording allows the sound engineer to use far more of the working range of a system than 16 bit systems ever could. In 16 bit environments maximizing the signal level into the A/D conversion stage is important to suppress system noise (yes, there is noise with digital…) In high-quality 24 bit conversion systems getting analog signals as close to 0 dBFS is far less critical from a noise perspective than with 16 bit. While best practice dictates using all of a systems dynamic range, when levels are set incorrectly the 24 bit system is far more accommodating than 16 bit."
 
You are 100% correct in describing the system as being top-fixed. What we do when we set audio levels is adjust the peak voltage hitting the A/D converter to coinside with the highest number available.

Think then this 1 bit for 6 dB of dynamic range the other way around: for a signal 6 dB more quiet than absolute allowed peak we loose the most significant bit which with binary numbers means division by 2. This is because the 6 dB more quiet signal has half the voltage.

1111 1111 1111 1111 = 0 dBFS
0111 1111 1111 1111 = -6 dBFS (this number is half in size of the above)

and so on, until we get to the situation where we have only zeros and we are at 16*-6= -96 dBFS. Here we hit the random digital noise if we have good enough analog signal.

With 24 bit system we can go 8 steps or 8*-6 = -48 dB further, which gives theoretical 144 dB dynamic range for 24 bits. It is theory only, as no analog system can give this kind of dynamic range in the feed signal.

Half amplitude = half the voltage = devide sample by 2 = most significant bit turns from 1 to 0.

This is where the 1 bit = 6 dB dynamic range comes from. With correctly set signals it is added to the quiet part of the signal.
 
You must log in or register to reply here.
Back
Top