At 09:56 AM 9/8/2002, you wrote:
>At 09:29 AM 9/8/2002, Bernie Krause wrote:
>
> >Does anyone know of a good match-to-sample software that is
> >available? Looking for something that could be conformed to identify
> >bird vox w/in the context of an entire biophony.
Doug wrote:
>I think this is the piece of software we all want - recordists as well as
>biologists who want to census an area, such as a marsh or a rain forest. In
>all the discussions and correspondence I have received, nobody has yet
>developed anything really workable. A few are working on it.
It occurs to me that Marty's new software could be used as the initial
"contextual filter" for a future comparator program. It might work like
this: The sound file is opened to the comparator program and the contextual
info (region, altitude, time of year, habitat, etc.) are entered into a
table. The comparator then needs to compare only those species that Marty's
program says are possibilities to the sample in question. This could reduce
the possibilities to less than a dozen in some cases. One could also enter
certain boolean parameters "manually", such as "Look only at Rails" or "Do
not look at Passerines".
The trick is to get the range of possibilities down to a reasonable number
before the actual search takes place. The program would also have to assign
a "confidence number" to each possibility on the list (Virginia Rail - 78%,
Sora - 22%, etc.).
Doug
Doug Von Gausig
Clarkdale, Arizona, USA
Moderator
Nature Recordists e-mail group
http://groups.yahoo.com/group/naturerecordists
[Non-text portions of this message have been removed]
________________________________________________________________________
________________________________________________________________________
|