American Birding Podcast



Documenting Rarities, Part 2: Have Fun, Get Fancy, and Learn Cool Stuff

At the end of Part 1, we listened to a pair of recordings of a Western Screech-Owl from Alamosa County, Colorado, on March 30, 2012. The owl’s hooting is faint in the first recording, and you can hear some amount of background “static.” In the second recording, however, the owl’s hooting is a fair bit louder, and the static is gone.

Hang on a second. There aren’t two recordings. We’ve been listening in fact to just one recording: We’ve been listening to the same bird, and the same series of hoots; I didn’t move around out there, and I didn’t adjust the settings on the recorder; again, we’re listening to two versions of the exact same recording.

Needless to say, I’ve “doctored” one version of the recording. In a nutshell, I applied a noise filter to the raw recording, then increased the amplitude. The result: a doctored recording with no static, and a louder owl.

How did I do that?


91 Eskimo CurlewsLong ago, folks would doctor photos with box cutters and glue. A famous example, discussed in detail by Martin Collinson in a recent article in Birding (September 2011 issue, pp. 26–31) involves photos of one or more Eskimo Curlews from Texas in the 1950s (see image, right). These days, we do it with computers. Download your digital photo of a common Cassin’s Vireo (or scan an old slide), mess around with the color and contrast settings in PhotoShop, and—voila!—you have yourself a rare Blue-headed Vireo.

I hasten to point out that photos can be digitally manipulated for less nefarious purposes—for example, to crop out a branch or twig, or perhaps to call attention to a particular feather or other field mark.

We can do the same sort of thing with audio recordings. No need for scissors and Scotch Tape. (Yes, I realize that readers under the age of 30 will have have no idea what I’m talking about.) Just download the soundfile (one word), create a sound spectrogram (I’ll explain in a second), and get to work.


If you’re still with me, you’re probably on a computer and you’re probably online. (People don’t really download and print this stuff out, do they?) Thus, you have access to one or more soundfiles. Here are two examples of how they might appear on your computer:


92x files       92y files

Open up one of the files, and most computers will play the sound for you. When I open up the 2.6-megabyte file called “Bushtit.wav,” I hear the sputtering calls of a male Bushtit, recorded in Boulder County, Colorado, on April 15, 2012 (we heard this same clip in Part 1). I also hear the annoying sound of a passing car.

And I think that’s as far as most of us ever get with listening to birds (and annoying background sounds) on our computers. Until recently, it’s as far as I ever got.

Let’s take it to the next level.


We need software that lets us “see” what we’re hearing. We have lots of options. I use a free application called Audacity, but you can use something else.

Now instead of opening the Bushtit soundfile with the default application for my computer (it happens to be iTunes for me), I’m going to open it with Audacity.

The result is utterly different. Instead of hearing the sound, I now see it. Here’s what I see:

93x Bushtit-raw
The result is a graph of frequency in kilohertz plotted against time in seconds. (Irksomely, Audacity places the time axis across the top of the graph.) Such graphs are often called “sonograms,” but I don’t favor that word. To me, a “sonogram” is a picture of a baby before it’s born. A sound spectrogram, then, is a plot of frequency against time.

And there’s something else on that sound spectrogram: lots of color. In particular, I see: (1) a mainly blue background with fine red dots; (2) a broad swath of red across the lower bottom of the graph (that is to say, in the lower frequency range); and (3) 27 vertical red stripes in the middle frequency range.

Those colors and patterns are a visual rendering of what we’re hearing. Let’s listen again.

Immediately, we hear the loud whooshing of a passing car, becoming fainter as the car gets farther away. That sound is represented graphically by the broad swath of red at the lower frequency range. It’s so loud, it simply drowns out the soft background “static,” represented on the graph by the blue wash with red stippling.

We also hear the notes of the Bushtit. Those notes are represented on the sound spectrogram by the 27 vertical stripes. They’re coming through pretty clearly, both aurally via the VN-8100PC and visually via Audacity, but, still, that passing car is awfully annoying.

So let’s get rid of the car!

In this doctored recording, the Bushtit sounds the same, but the sound of the car is greatly diminished. The car is noticeable—but bearable, I would say—for about the first two seconds, then softer and softer. I accomplished this trick via Audacity’s high pass filter. In geekspeak: cutoff frequency, 2.5 kHz; rolloff, 12 dB per octave; filter quality, 0.70. In plain English: I largely eliminated the low-frequency sound of the car, but preserved the relatively high-pitched calls of the Bushtit.


Bushtits are expanding their range north and east, and I wouldn’t be surprised if a small flock were to wander to the Nebraska panhandle. Suppose you were lucky enough to come upon one such flock. What would you do?

Well, you’d whip out your VN-8100PC, obtain a recording, and present the evidence before the Nebraska Ornithologists’ Union Records Committee.

“Sure sounds like a Bushtit,” you’d confidently note in your write-up on Nebraska’s first record for the species.

Hmm… What exactly does a Bushtit sound like? Does it say spit or pssst or ssip? How do such utterances differ from the calls of, say, Virginia’s or Orange-crowned warblers? How about the calls of even Long-billed Dowitchers or White-rumped Sandpipers? I’m serious: A whole lot of birds give monosyllabic call notes that can be rendered spit, pssst, ssip, or whatever.

This is a job for Audacity.

We’re going to “magnify” the sound spectrogram by a factor of 16. The procedure is extremely simple: We simply stretch out the time axis by a factor of 16. Also, I’m going to splice out a lot of the “space” (time, really) between calls. Here now is a clip from that spliced Bushtit recording, with the time axis elongated by a factor of 16:

93y Bushtit-magnified
There’s a lot of complexity in those 80-millisecond notes! Here’s what I’m seeing: a fairly faint vertical bar; then a louder (brighter) note that rises quickly from 4 kHz to 8 kHz; then a wavy descent to 5 kHz; then a rise to 7 kHz. Sure, there’s variability from call to call, but the basic pattern is consistent: faint vertical bar, then loud and rising, then descending and wavy, then rising a bit. To our human ears and brains, it’s merely spit or pssst or ssip. To a Bushit’s ears and brains, though, it’s a lot more. Don Kroodsma has written a lot about this—in the technical literature, in several popular books, and in this online tutorial for Birding.


I mentioned Orange-crowned Warbler, Virginia’s Warbler, Long-billed Dowitcher, and White-rumped Sandpiper. You could probably get away with spit or pssst or ssip for their calls. Let’s instead look at their calls, all recorded earlier this year in Boulder County, Colorado. Look how different they are from one another:

94a OCWa      94b ViWa

94c LBDo      94d-y WRSa

The Orange-crowned Warbler (top left, recorded May 4) is a V-shape. The Virginia’s Warbler (top right, recorded May 26)  is an S-shape, flipped, and lying on its side. The Long-billed Dowitcher (bottom left, recorded May 6) is a stack of four shallow crescents. And the White-rumped Sandpiper (bottom right, recorded June 11) is a shallow W, angled clockwise a bit, with a ghost of the same shape above. And there are other differences—in frequency, duration, amplitude, and so forth—all plainly evident from the sound spectrograms.


Let’s go back now to the matter of rare birds wandering to Nebraska. As you will recall, we were talking about a hypothetical flock of Bushtits. But let’s now entertain an alternative fantasy. Let’s pretend we’re dealing with a Black-capped Vireo.

Our hypothetical Black-capped Vireo is frustratingly silent, but it poses right out in the open. You get 150 frame-filler photos. Then, amazingly, the vireo flies a short distance to a nearby banding station, where it is captured, measured, weighed, photographed 150 more times, and let go. Next, the bird heads in the direction of the local nature center, with its big glass windows. Uh-oh. Well, you retrieve the specimen and deposit it with the University of Nebraska State Museum.

There’s just one thing left to do: Write up your report for the Nebraska Ornithologists’ Union Records Committee.

Now you do something bizarre. In your report, you neglect to include any of the hundreds of photos of the Black-capped Vireo; you decline to report any of the banding data; and there’s no mention in your report of the whereabouts of the specimen. Instead, you simply report that you watched the bird, and it sure looked like a Black-capped Vireo: black head, white goggles, the whole nine yards.

You wouldn’t do that, would you?

And, yet, I think we do it with bird sounds—even when we have recordings—all the time.

“Yep, sounds like a Bushtit to me. Sputtering, lisping, the whole nine yards.”

Don’t do that! Instead, make a sound spectrogram, and evaluate the actual data—just as you would for primary projection, feather wear, and maybe even DNA. If you ask me, the species-specific sound spectrogram of a Bushtit’s call is frankly the most distinctive thing about the bird! I mean, what else is there to a Bushtit?—that drab, colorless ball of cotton with a toothpick for a tail. Yet its sound spectrogram, we have seen, is possessed of wonderful complexity.

That’s enough, I think, on Bushtits.


I mentioned in Part 1 that we’d return in Part 2 to the matter of Willow Flycatchers. As you may recall from Part 1, I recorded a distant but definitive Willow Flycatcher on May 24 of this year in Boulder County, Colorado. The bird was—and, at this writing, still is—on territory in the lowlands in the eastern part of the county. So which subspecies is it?

(Before we go any further, I want to be clear that it is not my intent to get bogged down in Willow Flycatcher nomenclature. Suffice it to say, the topic is complex. If you simply must to know all the details, google the BNA Online account for Willow Flycatcher—but not until you’re done reading my words, please!)

Anyhow, is it an “eastern” or a “western” Willow Flycatcher? My honest answer to that question is: I don’t know. I’ve never seen this bird well enough to say whether it is greenish-brown (eastern?) vs. brownish-green (western?) above.

But check this out: I have a sound spectrogram, and recent publications suggest that geographic variation in Willow Flycatchers may have a critical vocal component. The sound spectrogram isn’t great, but I bet it’s good enough. Here it is:

95 WiFl
Again, it ain’t pretty. But there’s abundant detail in there. We can “see” that the fitz part of the fitz-bew song runs for about 140 milliseconds, during which time it rises from around 3.5 kHz to around 6.0 kHz. We can further say that the fitz part of the song has 11 discrete elements; and, if we wanted or needed to, we could say specific things about each of those discrete elements. And that’s just the fitz part of the song. The bew part, as you can see from the sound spectrogram, is substantially more complex.

When I listen to that bird’s song, whether in “real life” or here at the computer, I hear fitz-bew, and not much more. Similarly, when I view that bird in the field, I behold a greenish-brown (or brownish-green?) dot on a distant perch. But if I were to examine a great digital image (or a specimen), I would see much more than just a dot on a distant perch. By the same token, when I actually look at the sound spectrogram, I see vastly more than just fitz-bew.


If a Willow Flycatcher’s fitz-bew song can be said to be complex, then imagine how much more complex a Warbling Vireo’s song must be! A mnemonic I learned years ago goes like this: I can see you, I can seize you, I will squeeze you, till you SQUIRT! That worked wonderfully for the Warbling Vireos I grew up with back East.

Then I moved out West, where the Warbling Vireos say something entirely different. There’s a lazy, gruff quality about Warbling Vireos out West. They say something like this: ...uh?…yeah…what…EVER…dewd

Now how do you quantify those differences? A great approach is to look at sound spectrograms. Here’s a bird with an “eastern” song, recorded in the lowlands of eastern Boulder County, Colorado, on May 29, 2012:

96a WaVi-eastern
And here’s a bird with a “western” song (with warm-up by a Spotted Towhee), recorded a few miles to the west, in the Boulder County foothills, on May 25, 2012:

96b WaVi-western
(We heard these two birds in Part 1, by the way. Listen again to the “eastern” bird; and here’s the “western” bird again.)

I suspect you’ll agree with me that the images look different. Indeed, I can “see” eight potentially important distinctions between the two sound spectrograms. I don’t have the space here to go into all the details, but I lay it out for you in a recent tutorial at the Colorado Field Ornithologists’ website.


Warbling Vireos and Willow Flycatchers make me nervous.

Let me explain.

It wasn’t so long ago that most birders wouldn’t have been able to tell you the differences between what today many of us refer to as Eastern Warbling-Vireos and Western Warbling-Vireos. And it wasn’t all that long ago that birders knew only the “Traill’s Flycatcher”—which today we understand to consist of two species, the Willow and Alder flycatchers.

Fast forward to 2012. With new knowledge, anybody can separate Willow from Alder! Eastern campestris vs. western adastus?—No problem! As to a warbling-vireos out of range, just hear a snippet of song and just look at the clear differences in bill size!

If it’s not obvious, the preceding is a caricature. I’m being glib. But I bet you’ll agree with me that there’s a kernel of truth to that caricature. Don’t you know somebody who’s a bit too confident about separating invisible Setophaga warblers solely on the basis of their poorly heard nocturnal flight calls? Haven’t you ever been in a situation in which a single whit note was instantly and incredibly ascribed to this or that Empidonax flycatcher?

Have we gotten too cocky? Is it the fault of technology? We’ll explore those matters in Part 3.

Oh, and please don’t be misled or put off by those two questions I just posed. I’m not about to launch into some luddite’s diatribe against technology. Instead, I’ll be telling you about Alexander Pope, Jean Baudrillard, Rick Wright, and Red Crossbills. If that’s not an inducement to come back, I don’t know what is!