Hi Steve,
I would also claim that recording at 24 bit or 96 kHz does not have a signi=
ficant advantage for typical nature sound recording.
If you take a closer look at the samples posted here, you will find that th=
ere is usually very little energy above about 10 kHz. This also means that =
there is no reason to worry about sample rate conversion artifacts. These p=
otential artifacts are usually limited to signal frequencies close to the n=
yquist frequency (half of the sample rate). So, if here is almost no sound =
energy at those higher frequencies, this concern wouldn't be an issue.
It can also be seen (and heard) that the dynamic range of most nature recor=
dings (certainly soundscapes) is usually very limited. The dynamic range th=
at is present in those environments (except perhaps thunder) can be easily =
covered by a 16 bit recording system. An exception might be the (very low f=
requency) wind noise or microphone handling noise that can temporarily incr=
ease the demands regarding the dynamic range of the recorder. However, this=
would only be an issue if the microphone is not properly protected against=
these influences (lack of wind screen, high-pass filter and shock mount or=
pistol grip).
If one is interested in recording ultrasounds, then the higher sample rates=
are of course necessary.
> Does anyone know of other situations where signals above
> (say) 20 KHz are important?
Besides bats and many insects, rodents are known to emit ultrasounds. Certa=
inly the ultrasonic vocalizations of laboratory rats and mice have been stu=
died thoroughly (http://www.avisoft.com/rats.htm). The ultrasonic vocalizat=
ions of wild rodents are very similar (http://www.uncg.edu/~mckalcou/projec=
ts.html).
Recently, researchers have discovered that a few tropical frogs also emit u=
ltrasounds (http://www.plosone.org/article/info:doi/10.1371/journal.pone.00=
05413)
>Sonograms work fine but I haven't yet found a useful transformation
>and playback system that presents the sounds to other people.
The simplest way to make ultrasonic vocalizations audible is to play them b=
ack at a slower speed. This can simply be accomplished by modifying the sam=
ple rate entry in the .wav file header (don't confuse this method with re-s=
ampling). For instance, if you change the original sample rate from 192 kHz=
to 48 kHz, you would slow-down the recording by a factor of 4, which will =
also drop the pitch of the vocalizations accordingly.
Regards,
Raimund
|