Community Forums Archive

Go Back

Subject:24 bit/16 bit
Posted by: schirall
Date:5/11/2004 7:41:18 PM

what good is recording with 24 bit when you only have to mix down to 16 for cd's?

where are the 24 bit cd's?lol

Subject:RE: 24 bit/16 bit
Reply by: Iacobus
Date:5/11/2004 8:05:52 PM

I've noticed Best Buy has a 24-bit Super CD section now.

In any case, check out iZotope's guide to dithering. They can explain it a hell of a lot better than I can. :)

Iacobus
-------
RodelWorks - Original Music for the Unafraid
mD's ACIDplanet Page
Guitars 4 Kids

Subject:RE: 24 bit/16 bit
Reply by: schirall
Date:5/11/2004 8:09:47 PM

thanks. how is that ozone by zotope for mastering,ever use it?

Subject:RE: 24 bit/16 bit
Reply by: BarryBond
Date:5/12/2004 7:40:00 AM

I think the main reason for recording and mixing at 24 bit is because most plugins process at 24 and 32 bit resolutions. This way you get a much better resolution when recording and mixing. You can finally dither it down to 16 bit at the mastering process. This concept doesn't work the same way for sample rates unless it is a multiple of your final sample rate. Acceptable sample rates to work in when your final media is CD would be 44100hz, 88200hz, 132300hz, 176400hz, etc.

Subject:RE: 24 bit/16 bit
Reply by: schirall
Date:5/12/2004 4:31:02 PM

i gotta believe that 24 bit cd's are on the horizon,right. i hate mixing at 24 for an hour,then the last step is dithering it down.
it just ain't right,man,it just ain't right.

Subject:RE: 24 bit/16 bit
Reply by: BarryBond
Date:5/13/2004 6:09:03 AM

If anything's on the horizon it'll be DVDs. I bet eventually CDs will be phased out and all media will either be played on DVDs or portable harddrives/memory cards. Why do you hate mixing at 24 bit? Does it hurt your ears? I still don't see your point.

Subject:RE: 24 bit/16 bit
Reply by: pwppch
Date:5/13/2004 7:02:55 AM

Not even an issue. Most if not all audio editing/DAW tools do not work internally at 16 or 24 bit fixed point. The audio being used is converted to normalized 32bit floating point format before it is mixed or processed. I know of no current plugins that cannot handled this format. Some models can't handle anything but floating point.

So, regardless of the format of the actual media, all mixing and processing is done in floating point. The final stage of dithering is dependant on whether you are mixing in realtime to hardware, where the hardware bit depth is the controling factor or when you are rendering to a file (burning to a CD) where the rendering format defines the bit depth.

Peter

Subject:RE: 24 bit/16 bit
Reply by: schirall
Date:5/14/2004 3:22:14 AM

i don't hate mixing at 24 bit.i'd like to keep it at 24 bit.no dither,no adding noise no getting rid of the "least significant 8 bits"what i mix is what i'd like to play in the car,that's all i'm saying.why not?

Subject:RE: 24 bit/16 bit
Reply by: pwppch
Date:5/14/2004 6:59:38 AM

I have no idea what you are asking for.

There is _nothing_ that says you must dither. If you want a 24 bit result, just render to 24 bit when your project is done. There is no such thing as a 24 bit RedBook CD delivery medium. 16/44.1 is where CDs are at and this will not likely change. If you want to deliver in 24 bit, then you have to look to another medium.

Peter

Subject:RE: 24 bit/16 bit
Reply by: BarryBond
Date:5/14/2004 8:08:39 AM

I swing either way on this debate. I don't think you'd notice a difference if you recorded at 24bit and dithered to 16bit, recorded at 16bit, or recorded at 24bit and left it at that. The only way you may really notice is if you've got a $1million sound system. I only record in Protools and I do it at 24bit and dither to 16bit at the mastering process. I haven't noticed any problems yet.

I think the 24bit SACD is dead. I don't even think it took off. You need a special player to play them anyways, so why not make a superior audio format on DVDs. I bet more people have DVD players in their car than people have SACD players in their home.

Subject:RE: 24 bit/16 bit
Reply by: spinweb
Date:5/21/2004 12:22:10 PM

I just bought an ECHO I/O card that is 24 bit, where as the sound card that came with my laptop is 16 bit. If the file sizes of 24-bit are significantly higher and CDs themselves are only 16-bit, why bother with 24-bit sound (recording with it *and* rendering to it)? Especially if most folks render their files down to 16-bit in the end.

Subject:RE: 24 bit/16 bit
Reply by: BarryBond
Date:5/21/2004 2:05:17 PM

taken from http://homerecording.about.com/library/weekly/aa052698.htm

"The outcome of having more bit resolution is that the rises and falls in volume are much more natural because there is less of a jump between levels. To my ears, things like cymbal crashes, pops from an electric bass, drum hits or anything having transient peaks sound much smoother and more natural. The whole recording sounded more "analog" to me. That's a subjective statement but it's one of the best ways to explain it. My ears didn't tire as easily and I found the whole experience to be surprising and pleasing to my ear.

You may have heard that you should hit a digital machine as hard as you can to use as many bits as you can. What happens here is that at the lower volume levels the resolution is bad and your signal is overcome by quantization noise. The resolution is better up in the higher range. The problem comes when you have a signal with a large dynamic range and you're forced to record near those lower bits. The increased dynamic range of 20-bit or more represents the lower sounds stay out of the digital mud that's at the bottom of the bit ladder."



I also found a thread arguing both sides.

http://www.recording.org/modules.php?name=Forums&file=viewtopic&t=19153

Subject:RE: 24 bit/16 bit
Reply by: jardeano
Date:5/22/2004 10:15:11 AM

simply,, the higher the bit resolution the more the signal is represented ,as in pictures ,,, The conversion of analog to digital is complex and one has to take alot into consideration when expecting desired results, it's importantant for creating masters (higher bit and sample rate) because more of the file is represented, it allows less margin for error ,,lost data ,,clipping etc,, the original file has to be recorded at the higher bit resolution to reap its rewards, expert sound engineers like to record the sound with the effects they want allowing for a more workable file,(this method takes experience),to say what sounds better is where the discussion starts and debates begin,,,,,,,

Subject:RE: 24 bit/16 bit
Reply by: spesimen
Date:5/25/2004 2:51:11 PM

if you can't hear the difference when using dynamics processing on 24 bit files, then your monitoring setup isn't very good, your dac's aren't reproducing the sound well, or your ears are still a bit underdevleoped and ya need to keep at it!

now, if you compare a raw 24bit recording or a raw 16 bit recording that are made with good adc's, it is very difficult to tell them apart because the 96dB dynamic range of 16bits is close to what the ear can realistically perceive.

BUT...as soon as you start fiddling with the signal, particularly with compression types of effects, and when you start adding multiple tracks of that together, the difference is very noticeable. it's about the little bit of innaccuracy adding up over and over again. with 20-30 tracks that's a lot of little bits of innaccuracy! :)

in a nutshell, if you are just doing field recordings or whatever you're right that there isn't a huge need especially if you have a great mic/pre and adc. as soon as you are mixing multiple parts together, just go for the 24bits, you'll thank yourself for it later. back in the day when computers were slow and hard drives were small the storage and speed issues made this more of a problem but nowadays there's really no reason to settle for 16 in the early mixing stages.

cheerio

Subject:RE: 24 bit/16 bit
Reply by: pwppch
Date:5/25/2004 6:54:42 PM

>>BUT...as soon as you start fiddling with the signal, particularly with compression types of effects, and when you start adding multiple tracks of that together, the difference is very noticeable. it's about the little bit of innaccuracy adding up over and over again. with 20-30 tracks that's a lot of little bits of innaccuracy! :)
<<
This would be true if you mixed/processed in fixed point, but since most if not
all DAW type apps work in normalized floating point, what you are stating is just not the case.

Peter

Subject:RE: 24 bit/16 bit
Reply by: spesimen
Date:5/25/2004 7:35:20 PM

that is very interesting, i didn't know that. thanks for setting me straight.

it still seems like i can notice a big difference between the two, particularly with transient clarity on the low end. what else could cause that? psychology? it seems most noticeable with compression - perhaps my compression plugs are using an algorithm that isn't in normalized floating point? i switched to 24bit many years ago so perhaps the 16-bit apps back then weren't using normalized floats?

Subject:RE: 24 bit/16 bit
Reply by: spesimen
Date:5/25/2004 7:39:57 PM

also, why does acid (and most apps) support 24bit if it isn't needed? there is something gained from using it isn't there? isn't there some accuracy difference between 16 and 24 bit when the integer sample is converted into floating point at the first step? is it just that you don't have to worry about the decrease in dynamic range so problems with the original recorded signals not being hot enough are more easily ignored?

here is an example of what i was trying to say, comparing 4bits vs. 5 bits for hypothetical purposes. say there are 5 tracks being added together:

4bit case:
0001 (1)
0010 (2)
0110 (6)
0110 (6)
1101 (13)
total = 28/16 fullscale, (assuming some extra bits for adding them together, or we could divde by half like a master fader and get 14/16 in a purely int world)

5bit, converted straight from 4 bit:
00010 (2)
00100 (4)
01100 (12)
01100 (12)
11010 (26)
total= 56/32 full scale (or exactly double the above, which is identical)

5 bit, with a few differences on the lsb that may be introduced by adc:
00010 (2)
00101 (5)
01101 (13)
01100 (12)
11010 (26)
total = 58/32 full scale, so there is a slight difference when just adding ints..(would be 14.5 when divided back down to fit the 4bit scale, so it would probably round one way or another if a float was used, but be 14 if you just drop the bits off by shifting them right?)

now if it’s converted to float, and this is where I’m not really understanding what happens, wouldn’t the float representation of the set in the third group still be different than the second group? even if the accuracy of any subsequent operations is identical from happening in floating point, the inital values that are represented are still more accurate aren't they? and perhaps subsequent operations (especially a lot of them like a big plugin chain) would be more accurate? or is that irrelevant?

anyway i would prefer to avoid putting my foot in my mouth again on this topic so any info would be very helpful.

Subject:RE: 24 bit/16 bit
Reply by: pwppch
Date:5/26/2004 12:29:17 PM

The bit depth you set is for hardware i/o only - including rendering to a final file. Yes, there is something gained - if your hardware has 24 bit converters that exploit this. More bits give you more dynamic range. Softer softs and louder louds. However, depending on the abient of your room and the quality of your converters, the overall benifit of more 'bits" varies.

Normalization means that 0 dB in a 8, 16, 24 bit file is the same : 1.0f. Same goes for other points on the dB scale : - 6db is represented exactly the same for fixed media at any bit depth when converted to floating point.

If your fixed point media has 24 bits of dynamic range, then you will have more 'information" to work with. As you start to mix and process, you could either expand or decrease your dynamic range.

Normalized floating point has a "built in accumulator". No need to carry around and accumulator for mixes that go over the fixed point bit depth.
The math is fast on modern CPUs and you can count on nearly all plugins/FX to support it. That is you don't have to worry about how different plugins handle the "accumulator".

Processing - addition and subtraction - are easier using floating point. Creating a good mix still means that you have to pay attention to the max in the digital world of 0 dB. If you mix hot > 0 dB , you will still get clipping.

Peter


Subject:RE: 24 bit/16 bit
Reply by: spesimen
Date:5/26/2004 4:41:20 PM

thanks for the insight!

Go Back