At 7:13 PM +0000 5/13/07, Raimund Specht wrote:
>Yes, David is right. I agree that the editing procedure should be done
>at a higher resolution than 16 bit. But that does not mean that you
>need to initailly record in 24 bit! The format conversion can simply
>be done before starting the editing procedure (most audio editing
>applications do that automatically).
>
>We naturerecordists usually deal with relatively low dynamic ranges
>that can be easily covered by the 16 bit format. Even if you record a
>loudly singing bird at a close distance, the sound level at the
>microphone will usually never reach 80 dB. Then subtrcat the self
>noise level of the microphone (e.g. 5 dBA if you have a good very good
>one) and you will get a theoretic dynamic range of about 75dB only.
>So, there is still enough head and foot room within the theoretic
>dynamic range of the 16 bit format (96dB). Please note that I still
>have not included the background noise (such as traffic noise, wind
>noise and so on) into that calculation...
>
>Raimund
I agree that nature recordists can often record at a higher levels to
take advantage of limited dynamic range but, sometimes, the location
sound levels result in very low bit depth even with the recorder's
pre is at full gain. It was interesting for me to note that the sound
levels used in Sound Device's convincing test comparing 16 and 24 bit
recording are on par with the saturation I get from ambience in
remote settings in the middle of the night with a mkh40-> 744T at
full gain (-40dB peaks). I edited the low saturation (but
normalized) -40dB 16 bit and 24 bit recordings together into one file
so one can assess the noise differences better:
.mp3/256kbs [.5mb]
http://www.uwm.edu/~type/audio-reports/LowSaturation/AudioClips/16vs24bit-40dBCompare.mp3
.mov/H263_IMA:4 sound [1mb]
http://www.uwm.edu/~type/audio-reports/LowSaturation/AudioClips/16vs24bit-40dBCompH263_IMA4.mov
Thanks for pointing us to the excellent SD test, Tim!
Rob D.
|