American Birding Podcast



How to Record Birdsong—Part 2

Note: This is the second in a three-part series on how to record bird vocalizations. This part is about software—editing sound files and then sharing them with other birders on the internet. Click here to back up to the first part—an overview of the hardware you’ll need to get started.


5. Edit the sound file.
 I’ll start off on a heretical note: I use Audacity, not Raven, for editing and displaying sound files. I accept that Raven is the industry standard for birding, but Audacity is better suited to my needs. Files edited in Audacity sound superior to my ears—smoother, less harsh, than files edited in Raven. And I actually like the parti-colored spectrograms generated by Audacity; they help me discern differences in “volume” (technically, millipascals of sound energy) in a bird’s song or call.

You don’t have to agree with me on this. Submit a comment, below, if you don’t. Let’s move on.

First, I ought to mention three glitches with Audacity. On my Mac, at least, the program occasionally crashes (no biggie, just re-launch), occasionally fails to open (more annoying, I have to reboot), and very rarely disables all sound on the computer (a headache, resulting in my having to reset the computer’s PRAM, which google).

Now an item of housekeeping: Always keep the original, unedited, unaltered, undoctored, “raw” sound file. There are two reasons to do this. First, you never know when there will be something of interest lurking somewhere in the sound file; I’ve several times had the experience of discovering something notable only after going back and listening to the recording on the computer. Second, if the sound file constitutes proof of a rare or otherwise notable bird, you want evidence that hasn’t been tampered with. It’s the same way with a photo of a rare bird: Sure, it’s fine to clean it up (sharpen the contrast or bring out the earth tones, crop out an ugly building or clone out a twig, etc.), but the records committee most likely wants the original, unaltered image.

We’ll continue now with a real example, a Lark Sparrow I recorded on Friday afternoon, May 23, 2014 in Las Animas County, Colorado. Here’s the sound spectrogram of the entire, unedited file:

09-Lark Sparrow spectrogram

The first thing to notice is that I’m showing just a single track (or “channel”), whereas the Audacity output shows two tracks. What’s going on? Since I’m using the zoom (“shotgun” or “mono”) option on the LS-10, the two tracks are identical. That’s good. It means I’m focused on the bird, as opposed to surrounding sounds—of which there were several. (Google “cardioid microphone,” if you want to get technical about it.) Another advantage of the zoom setting is that it tremendously simplifies spectrographic analysis of a recording: You get all the signal (all the “information”) in one track, as opposed to a complicated function of partial signals in two places.

Let’s get back to our Lark Sparrow.

The bird starts singing a bit before 11 seconds and continues to a bit after 14 seconds. At around 16 seconds, I start to walk toward the bird; you can “see” my footprints in the form of those tall red columns. Next, the bird flushes, and I start talking into the recorder a bit after 19 seconds.

I’m not interested in my footsteps or voice, so I delete all but the bird. Well, almost all. It’s good to leave some space (or, rather, time) before and after the vocalization. So we get this:

10-Lark Sparrow spectrogram

All those squiggles are the song of the Lark Sparrow. But do you notice something else? That strong band in the low-frequency range is the wind—which would be even more noticeable with no wind baffles. The wind diminishes the quality of the recording.

Audacity to the rescue: Simply apply the HIGH PASS FILTER with a not-too-severe ROLLOFF. You’ll get good at this after you play around with it for a while. For now, take it on faith that this recording will benefit from a cutoff frequency of 1,500 Hz and a rolloff of 24 decibels (dB). The wind is gone:

11-Lark Sparrow spectrogram

Next we take a look at the waveform function of this bird’s sound:

12-Lark Sparrow waveform

There are two things to notice. First, this is a clean—that is to say, cleaned-up—recording, indicated by the nearly perfectly thin trace leading up to and trailing the song. With the wind gone, there’s practically no background noise. The second thing to notice is that the bird itself isn’t particularly loud: The waves’ amplitudes aren’t all that great.

The reason the bird isn’t all that loud is because I wasn’t all that close to it. In the photo below, I’m in the process of actually recording the very Lark Sparrow under consideration right now. As you can see—or, I guess not see—the bird is well outside the photo:

13-Ted Floyd with recorder

Photo by © Chris Pague.

What to do in situations like this? Well, if we were looking at the real bird in the field, we would raise our binoculars. Or if we had a high-resolution photo, we would zoom in. We can do something analogous by applying Audacity’s AMPLIFY feature. Go easy here. I almost never crank it all the way up. In fact, I usually don’t amplify at all. But we’ll do it on this somewhat distant Lark Sparrow, and we’ll go easy—5 dB, instead of the maximum allowable 15. Here’s the result:

14-Lark Sparrow waveform

You can see the change. The bird is a fair bit louder (the waves have greater amplitude), but there’s more background noise (the thin trace before and after isn’t as thin). Ratcheting up the “volume” increases the “static.” It’s an unavoidable tradeoff, analogous to increasing the magnification on a zoom scope: bigger bird, poorer image. Here’s the same result, shown a different way:

15-Lark Sparrow spectrogram

Compare it with the previous spectrogram. The bird is louder now (brighter traces on the spectrogram), but there’s more background noise (heavier broadband red–purple spackling).

You might want to know what the bird sounds like, yes? Here you go:

I think the final result is decent, but not great. The bird was at some distance, and it was a windy afternoon. In this recording, you can certainly hear that it’s a Lark Sparrow. And you can “see,” by way of the spectrogram, various features of the bird’s song: how some notes fall in pitch and others rise, the exact number of notes in the different trilled elements in the song, the changes in volume throughout the song, and so forth.

This is just the tip of the iceberg. With Audacity, you can do a whole lot more than clip out footsteps, eliminate wind, and amplify birdsong. Audacity is Photoshop for audio recordings. Which brings me to these two key points:

  • By all means, use Audacity—and all of its cool features—to learn about the quantitative features of bird vocalizations: basic stuff like duration, loudness, and pitch, as well as “advanced” stuff like modulation, intonation, and nasality. By playing around with the pitch, speed, and tempo settings in Audacity, I’ve learned amazing stuff about the unbelievably rich songs—plural, songs—of the House Sparrow.
  • That said, a doctored file is, well, a doctored file. It’s difficult and often impossible to undoctor a doctored file. So hang onto the original, unaltered file—ideally, in “raw” .wav format. Better yet, obtain a recording that doesn’t require doctoring. I’m going to come back to that point when I finally get around to step #2. But we’re not quite there yet.


6. Upload to the internet. You’ve just recorded a cool bird—a rarity, let’s say, or “just” a beautiful song, or perhaps an interesting or aberrant vocalization. Now what?

Not too long ago at all, your only option was that proverbial shoebox in your mom’s attic. You’d toss the reel-to-reel tape in a crate or carton, and soon forget about it. One of the great birdsong recordists—an older guy—tells me that he’s never once listened to the majority of his recordings. They’re on reel-to-reel, uncataloged and uncurated, in unopened boxes in an archives in another state.

Life before the internet was so barbaric.

A quick and dirty way of getting your sound files out there is SoundCloud—easy, fast, and annoying.

Let’s deal first with “easy” and “fast.” Back on June 17, 2014, my kids, Hannah Floyd and Andrew Floyd, and I were looking for the pleasing fungus beetle Gibbifer californicus. But you’re never not a birder, the saying goes, and when we heard a Lazuli Bunting, Andrew recorded its song. Here is Andrew’s recording, via SoundCloud:

The upload to SoundCloud requires just a few clicks. (Note, by the way, that Andrew provides a bit of commentary at the end—a nice touch. You will recall from step #3, “Operating the Hardware,” that it’s good practice to speak a few notes into the recorder.)

Now for “annoying.” SoundCloud tries to read your mind—and does a really bad job of it. Assuming you’re not logged into my account, SoundCloud suggests that you listen to other recordings uploaded by me—recordings that aren’t all that interesting. And if you are logged into my account (or, more likely, logged into your account), it gets worse: the dreaded SoundCloud loop. I swear, it’s like something out of Kurt Vonnegut’s “Harrison Bergeron.” Does anybody know how to disable this annoying feature of SoundCloud?

As to file management, SoundCloud is irredeemably hopeless.

So let’s go for the gold. Instead of SoundCloud, we’ll use Xeno-Canto—one of the most amazing resources in the whole history of birding. Getting a recording onto Xeno-Canto is a bit of slog, but it’s well worth the effort.

Xeno-Canto requires that uploads be of MP3 files. So that means converting from .wav to .mp3 format by means of a third-party converter like this one:

Now that the file is in the proper format, you’re ready to upload to Xeno-Canto. Be prepared for a full-on inquisition: Where were you? What was the date? What time was it? Did you see the bird? Did you use playback? Are other species audible in the recording? How many notes are in the bird’s song? Does the song rise or fall in pitch? Does it increase or decrease in volume? And a whole lot more. In truth, you can decline to answer a fair number of those questions; but I think it’s good discipline to go through the entire battery of questions. Speaking for myself, the Xeno-Canto inquisition has made me a better—more aware, more knowledgeable, more careful—hearer of bird vocalizations.

At last, the song has been uploaded:

This is the same bunting we clicked on a moment ago, SoundCloud-wise. Along with the recording itself, Xeno-Canto now provides a ton of information about the recording: the technical details of the recording itself, various quantitative and qualitative details about the bird’s song, and potentially important information about the bird’s acoustic and biological environment.

Xeno-Canto’s search engine—fast, powerful, and accurate—is almost too good to be true. Play around with it, and see for yourself. If you want to start with baby steps, try my personal page at Xeno-Canto:

Let’s get back to the Lazuli Bunting.

Something you can “see” in the 5–6 kHz band in the Xeno-Canto recording is crickets; you can also “see” broadband noise courtesy of a family group of Black-billed Magpies. It is possible—although you really need to know what you’re doing—to get rid of the crickets and magpies with Audacity. But there are two reasons not to. First, the crickets and magpies were there. They were part of the aural experience. It’s inauthentic, if you ask me, to eliminate them from the soundscape. Second, it might well have been possible to obtain a recording without crickets and magpies to begin with.

Which finally brings me to step #2.


Click here to continue reading.