Advantages To Using 24-Bit?

  • Thread starter Thread starter The Engineer
  • Start date Start date
T

The Engineer

New member
Can any one name some reason's why? I know it's the best bitrate to use, cause I'd like to gain knowledge on exactly why it's better. Also, does using 24-bit compared to 16-bit give you more headroom in recording? Just wondering.

One more thing, I know 32-bit isn't a "true" 32-bit, sooo if you record in 32-bit what bitrate is it ACTUALLY at?


Thank you to all who respond.
 
As I scanned the linked thread above I shuddered at some of the responses, so here's a few quick answers. I recommend picking up a copy of Pohlmann's 'Priniciples of Digital Audio'...

The Engineer said:
Can any one name some reason's why?
Click to expand...
Greater dynamic range.

does using 24-bit compared to 16-bit give you more headroom in recording?
Click to expand...
No, the greater dynamic range allows more 'foot room', so to speak. You are lowering the noise floor by 48 dB by going from 16 to 24 bit - maximum level is always 0dBFS in the world of digital. Of course, the finest converters made only give you about -120dBFS noise, so 20 bit is about as good as it gets in the real world.

One more thing, I know 32-bit isn't a "true" 32-bit, sooo if you record in 32-bit what bitrate is it ACTUALLY at?
Click to expand...
It's 24 bit that 'floats'. You actually get somewhere in the neighborhood of 1500dB dynamic range with 32 float.
 
Thanks Blackwood. Well I used to have nuendo and I was impatient to learn everything cause it was kind of confusing and since I was already used to Cool Edit Pro I just stuck with that. Though CEP doesn't have 24-bit recording. So, it would be better to record in 32-bit instead of 16-bit?

P.S. I'll eventually have to learn Nuendo, so I can record 24-bit. :)
 
The Engineer said:
Thanks Blackwood. Well I used to have nuendo and I was impatient to learn everything cause it was kind of confusing and since I was already used to Cool Edit Pro I just stuck with that. Though CEP doesn't have 24-bit recording. So, it would be better to record in 32-bit instead of 16-bit?

P.S. I'll eventually have to learn Nuendo, so I can record 24-bit. :)
Click to expand...

It would be better to record at 32 bit float than 16 bit all other things remaining equal.

32-bit IEEE-752 floating point format allows 24 bits of mantissa (where the non-exponent data is stored) and 8 bits for the exponent. So it will be more accurate than 16 bit fixed (more discrete levels with regards to the amplitude of the signal).

Some people say that they prefer 32 bit float over 24 bit fixed, however there may be issues in regards to to "quantization noise" that make 24 bit better, See:

http://www.bores.com/courses/intro/chips/6_precis.htm
 
Last edited:
24-bit resolution/depth actually makes a much bigger difference subjectively to my ears than 44.1khz or 48khz sampling compared to 96khz. What I mean is - use 24-bit where you have it in the coverters.

Tom - By 32bit float, do you mean what the software calles its bit depth?
 
Sorry, I just want to say that that's the worst post I think I've ever written. I'm a little sleep-deprived so I won't change it now, but feel free to step over it!!! :):)
 
noisedude said:
24-bit resolution/depth actually makes a much bigger difference subjectively to my ears than 44.1khz or 48khz sampling compared to 96khz. What I mean is - use 24-bit where you have it in the coverters.

Tom - By 32bit float, do you mean what the software calles its bit depth?
Click to expand...

Yes, the bit depth that it stores and processes digital audio.
 
noisedude said:
24-bit resolution/depth
Click to expand...
FWIW, it's popular to call the bit depth 'resolution' though the only added 'resolution' is below the LSB of the comparitive bit depth. For example, the top 16 MSBs in a 24 bit word are exactly the same as those of a 16 bit word - the extra 'resolution' occurs below -96dBFS. Because of this, I consider it incorrect/misleading to refer to bit depth as resolution...
 
bblackwood said:
No, the greater dynamic range allows more 'foot room', so to speak. You are lowering the noise floor by 48 dB by going from 16 to 24 bit - maximum level is always 0dBFS in the world of digital. Of course, the finest converters made only give you about -120dBFS noise, so 20 bit is about as good as it gets in the real world.
Click to expand...

I was trying to think of a way to explain that when I first saw this post, couldn't really put it into words, and you did it perfectly. This post is golden. :cool:
 
bblackwood said:
FWIW, it's popular to call the bit depth 'resolution' though the only added 'resolution' is below the LSB of the comparitive bit depth. For example, the top 16 MSBs in a 24 bit word are exactly the same as those of a 16 bit word - the extra 'resolution' occurs below -96dBFS. Because of this, I consider it incorrect/misleading to refer to bit depth as resolution...
Click to expand...

I like to think of it as more fractional values for each sample, so there is more accuracy involved during calculations (less quantization noise). Brad is correct, these additional fractional values are below -96dbFS.

This article explains it pretty well:
http://www.indiana.edu/~emusic/digital_audio.html
 
I apologise again for the incoherence of the previous post!! Yep, I understand exactly how the hardware process works - and so why resolution is a misleading word - but I have always been a little bit confused by the idea that my software could be recording at 16- or 32-bit through my 24-bit converters!!!

Thanks Tom and Brad. :)
 
You must log in or register to reply here.
Back
Top