Hi all,
Ug, one of those days: I realize I sent this reply directly to Eric
rather than the list.
-Mike
-- Forwarded message --
From: Mike Powers
Date: Fri, Aug 21, 2009 at 11:33 AM
Subject: Re: [nfc-l] detection settings for raven pro?
To: e kent
Hi Eric (and everyone),
We've been using Raven Pro 1.4 beta to scour our nocturnal recordings
for flight calls. I've been testing a new recording unit and running
Raven detectors on the sound files (WAV format) each morning. This
season I have found two detector settings that I'm overall very
pleased with.
Our best results (thus far) are from running two Band-limited Energy
Detectors (BLED), one that covers the 6 - 11kHz (mimicking Bill's
tseep detector) and one that is limited to 2.25 - 3.75 kHz (mimicking
Bill's thrush detector). I've attached both detectors to this email
if you (or anyone else) is interested in trying them, at least as a
starting point for your own recordings. If anyone does I'd love to
hear how they work and/or what adjustments were needed. I also wrote
out the settings below if you just want to peruse them.
A caveat: these detectors seem to work well for my site in
Horseheads, NY (35 miles southwest of Ithaca), but from our anecdotal
experiences it seems detectors may have to be tweaked to be
site-specific. Also, an exclusion band is desirable on rainy nights,
I've had good results using this feature this season. I included
these settings as well though I only use them when Raven returns over
1000 events (in the Sparrow band) due to rain. I'm still testing
Exclusion Band settings in the Thrush band to counter rain.
Bird wise: I found good numbers two nights ago, the night of 19
August. My Sparrow detector picked found 145 selections, 136 were
actual calls. I haven't gone back to put IDs and numbers on them yet,
but White-throated Sparrow, Ovenbird, American Redstart, Canada,
Black-and-white, and Black-throated Blue Warblers stuck out, there are
many Zeeps, Double-banded Up- and Down-sweeps that need closer
attention. I'm not done with the Thrush band yet, most (if not all)
of what I've seen so far are Veery. Apologies I don't have the IDs
and numbers broken out yet.
Again, the current settings are below my sig file, note that I did not
enter anything to four or five decimal places. Raven appropriately
re-sizes these entries automatically.
Cheers,
Mike
--
Mike Powers
Research Biologist, Conservation Science
Cornell Lab of Ornithology
159 Sapsucker Woods Road
Ithaca, NY 14850
email: me...@cornell.edu
SPARROW DETECTOR
Band Tab
Min freq: 6000 Hz
Max freq: 11000 Hz
Min Duration: 26.7 ms
Max Duration: 400 ms
Min Separation: 101. ms
Noise Tab
Min Occupancy: 25.0 %
Return Occupancy - unchecked
SNR Threshold: 25 dB
Block Size: 4997. ms
Hope Size: 245. ms
Percentile: 50.0
Exclusion Band unused, except on particularly rainy nights. If needed I use:
Min Freq: 2000 Hz
Max Freq: 4000 Hz
SNR Threshold: 6.0 or 8.0 (depends on how each fairs, but these seem
to give decent results)
Bandwidth Tab is unused.
THRUSH DETECTOR
Band Tab
Min freq: 2250 Hz
Max freq: 3750 Hz
Min Duration: 32 ms
Max Duration: 330.7 ms
Min Separation: 48 ms
Noise Tab
Min Occupancy: 20.0 %
Return Occupancy: unchecked
SNR Threshold: 4.0 dB
Block Size: 997. ms
Hope Size: 250.7 ms
Percentile: 50.0
Exclusion Band unused, except on particularly rainy nights and
selection tables are greater than 7500. I'm still working with this
to get a reasonable result.
Bandwidth Tab is unused.
Cheers,
Mike
--
Mike Powers
Research Biologist, Conservation Science
Cornell Lab of Ornithology
159 Sapsucker Woods Road
Ithaca, NY 14850
email: me...@cornell.edu
On Thu, Aug 20, 2009 at 6:29 PM, e kent wrote:
> Last spring I used the software tools from oldbird.org (tseep-x, thrush-x
> and GlassOfFire) to examine sound files. Unfortunately, it was a bit
> tedious as I was recording in an unsupported bit-rate. So I made batch and
> shell scripts to move, convert, and examine the files
> ~relatively~ automatically. However, I have Raven Pro now (1.4. beta) and
> am wondering if anyone has detection settings to share? That way, I can do
> all detections without spending time converting, etc.
>
> By the way, once I used "GlassofFire" I really liked how it lets you quickly
> categorizing and move results. Guess I'll see if it's still useful if I can
> move to doing all detections in Raven.
>
> Thanks for any help!
> Eric
>
--
NFC-L List Info:
http://www.NortheastBirding.com/NFC_WELCOME
http://www.NortheastBirding.com/NFC_RULES
http://www.mail-archive.com/nfc-l@cornell.edu/maillist.html
--
MEP-Sparrow
Description: Binary data
MEP-Thrush
Description: Binary data
