naturerecordists
[Top] [All Lists]

3. Re: 24 bit vs. 16 bit

Subject: 3. Re: 24 bit vs. 16 bit
From: "Rob Danielson" danielson_rob
Date: Wed Jul 25, 2007 8:02 pm ((PDT))
At 4:42 PM -0500 7/25/07, Bruce wrote:
>I wanted to put my two cents into the 24-bit discussion.
>
>I think you are missing the real advantages of 24-bit over 16-bit for
>recording soundscapes. I see it as an advantage to record at lower levels
>than I would have using a 16-bit system. When recording nature sounds there
>is always the unexpected sound that is louder than we would expect. To make
>best use of a 16-bit system, you would have to play with gain controls to
>achieve a signal of less than 0dB, but as close as you can to 0dB with out
>going over. You can be just happy with a setting, recording at -10 to -20db
>average, giving you around a 75dB dynamic range. Then a bird, a thunder
>clap, or other sound suddenly comes and clips the recording sequence, many
>times ruining the entire recording. Sometimes just a turn in the sound
>maker's body can clip the signal. With a 24-bit system, you can comfortably
>record at signal levels of -30 to -40dB average, while still getting over
>80dB of dynamic range. At those levels you are unlikely to have an inherent
>sound go into clipping, and gives you the added advantage of getting 
>rare sounds when say an owl perches over your mic set-up. Post 
>editing can bring the signals up to the level you want. I see 24-bit 
>systems as an advantage
>to the recordist to use the added dynamic range, not necessarily as being
>better sound quality. With my 24-bit system, I set the gain to 30, 40, or
>50dB and walk away. No fussing around anymore, watching meters. I have
>colored LED meters I can watch through binoculars from a long distance away,
>just to make sure everything is cool. I will say that I found out how easily
>a signal can saturate, and how many mini-disc type level meters are too slow
>to know the event occurred. Any saturating, no mater how brief will cause a
>loss in quality, and that alone maybe the advantage of sound quality in a
>24-bit system. While out recording this Spring I can't tell you how many
>times my recordings would have been ruined by some birds that came in close
>to my recording set-up if I was still using a 16-bit system and flirting
>with saturation. Instead, those recordings came out perfect. Very few 24-bit
>recorders can achieve true 24-bit performance. The only ones that I know of
>are from Core Sound and Sound Devices, but their specifications are more
>based on part specification than actual measurements from completed units,
>but they seem to work ok. The one problem is that those recorders have
>higher noise levels at low gain settings, making the whole advantage
>somewhat pointless. All those other "24-bit" recorders from Zoom, M-Audio,
>Fostex, and others should be thought of as 16-bit recorders and no more. The
>advantage of the M-Audio is that it at least has a 24-bit input to bypass
>their poor inputs. I still use 16-bit recorders for capturing singles, as I
>can adjust gain on the fly to get the best results as I follow the subject
>around.
>
>Bruce Rutkoski
>www.natureguystudio.com
>

Hi Bruce--
I hear you about the potential advantages with dynamic range.  My 
test/report is intended as a response to Sound Devices' page  
http://www.sounddevices.com/tech/24-bit.htm  suggesting that 24 bit 
recording offers less quantization noise than 16 bit recording under 
practical, "real world" recording conditions.

Though amplitude quantizing noise is the only performance factor I 
tried to study, one could deduce from the results that the dynamic 
range with 16 bit recording might be more than sufficient for many, 
many natural settings. My listening-based test suggests to me that if 
I create a _16 bit_ file with a background level above (louder than) 
-55 to -60 dBFS, I should be able to safely avoid any quantizing 
noise. I suppose one can measure the background levels in one's 16 
bit recordings after the fact for another sense of how much 
"under-recording" in the field one can get by with. SD did produce 
the quantizing noise in their test by "under recording" a very 
significant amount. I tend to not under-record to great extent, but 
should the circumstance call for lots of headroom, I'll be much less 
hesitant to do it in the future-- even with 16 bits. From a small 
amount of testing I've attempted with Richard's help, large digital 
gain increases  (like 30+dB) seem capable of altering qualities in a 
recording so one might want to factor this into one's gain strategy 
as well. Rob D.







<Prev in Thread] Current Thread [Next in Thread>
  • 3. Re: 24 bit vs. 16 bit, Rob Danielson <=
Admin

The University of NSW School of Computer and Engineering takes no responsibility for the contents of this archive. It is purely a compilation of material sent by many people to the naturerecordists mailing list. It has not been checked for accuracy nor its content verified in any way. If you wish to get material removed from the archive or have other queries about the archive e-mail Andrew Taylor at this address: andrewt@cse.unsw.EDU.AU