I guess the answer depends on the other details of the system for making th=
e recording and whether there is reason to believe that there is any inform=
ation in the higher frequencies. Certainly if one includes a strong filter =
cutting everything above 22 KHz (or whatever) then there's no point in samp=
ling faster then 44 KHz. If the microphones used are able to respond to any=
thing above 22 KHz, the same conclusion would apply, it seems to me,
There are some commercially available microphones that do work pretty well =
up to 40 or 60 KHz and with them one could make the case that it is worth s=
ampling fast enough to acquire the signal. Even though we can't play it bac=
k easily or effectively these days someone might want to do signal processi=
ng or measurements on the data in the future.
Data storage costs for me aren't too large compared to time, transport, etc=
. so I sometimes sample faster than 44/48 KHz on the chance it'll be useful=
to someone someday. For people who record more, storage and backup can be =
a significant expense.
I'd also like to point out (a topic not often covered here but related to C=
urt's question) that the question of archiving recordings so that future re=
searchers can learn about and get access to them and get the associated dat=
a is a significant one. The current best approach I know of is to donate or=
make plans to donate to one of the acoustics archives associated with a u=
niversity or museum. Many have long-term plans to protect their data. In th=
is context of future researchers, detailed field notes about the recording =
might be even more valuable than the 2x or 4x samples. I've often though o=
f adopting a scheme to put data =3Dpossibly large bits of text, photos, oth=
er documentation- into the sound file itself. Something like RIFF would let=
us do that, I guess.
Cheers!
Steve P
--- In Curt Olson <> wrote:
>
> We've discussed sample rates here before. If I remember right, some
> people claim they can hear a difference between 44.1/48 and 88.2/96 or =
> 192, but others say no.
>
> My question is not about that, but about the future. In cases where we =
> might be documenting soundscape data for future reference, do you all =
> think we should we make an effort to capture the highest possible
> sample rates -- just because we can? Or can we be confident we're
> serving future users well with 44.1/48 recordings?
>
> Curt Olson
>
|