naturerecordists
[Top] [All Lists]

Re: xeno-canto developments

Subject: Re: xeno-canto developments
From: "Peter Shute" pshute2
Date: Mon Jan 21, 2013 5:12 pm ((PST))
That sounds good to me, and not necessarily so hard.

Currently a file is associated with a species name (or several?). It needs =
to also (optionally) be associated with two times and two frequencies, thes=
e being the start and end of the call, and the upper and lower limits of it=
s frequency range (or the loudest parts). Raven Pro does this, I believe, a=
nd probably others.

Once the search has returned the results, it should be relatively simply to=
 use them to jump to that part of a track, and to display a rectangle on it=
s spectrogram.

Creating a web interface to allow entry of these ranges might not be so str=
aight forward, but as you say, it's all been done before with Facebook.

But is a rectangle good enough? On ambient recordings, sometimes the calls =
are so overwritten by multiple species and individuals that more precise id=
entification might be necessary, e.g. arrows and polygons.

This might be overkill. It's a big clue to the listener to at least have th=
e time and frequency range identified.

Peter Shute

> -----Original Message-----
> From: 
>  On Behalf Of Elias Elias
> Sent: Tuesday, 22 January 2013 11:39 AM
> To: 
> Subject: Re: [Nature Recordists] xeno-canto developments
>
> Regarding the idea of "drawing a box around a signal," I
> suppose that I meant  that it would be nice to tag noises in
> an audio file (long or
> short) much like one tags friends in a Facebook image.
> Essentially I envision dragging a courser (a cream-colored
> courser of course) around a sound (a frog's 'peep' or a
> sandpiper's 'churt' or an owl's 'who's awake? me too') and
> being able to fill in metadata such as species, sex, age,
> behavioral context and other parameters that XC already
> allows. Of course, the file will already have been date,
> time, latitude, longitude, elevation stamped. So say if one
> wanted to search for call notes of long-billed murrelet (I
> know there aren't any), one put that in the search box and
> bam up come up 25,50,100 thumbnails of sonogram images scaled
> to just the call or call phrase (these images would perhaps
> be squares much like google image search displays their
> search results). When the courser hovers over the image it
> expands to full resolution. With all the metadata in a table
> off to one side. If one clicks the sonogram it plays just the call.
>
> Dream dream dream....





<Prev in Thread] Current Thread [Next in Thread>
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