I have delved into
measuring different aspects of Licoln’s sparrow songs. I do all of these
measurements on a visual representation of sound, called a spectrogram. The
spectrogram represents the frequency of sound on the vertical axis and time
on the horizontal axis. The intensity of the spectrogram’s color represents its
amplitude.
 |
| Bird song represented as a spectrogram. Frequency (pitch) is in kHz, time is in seconds. The intensity represents the amplitude of the song (photo credit: Langmore 1998). |
The analysis of Lincoln's sparrow song first involves breaking the songs up into their
component parts, which we call syllables, (and trills when the same syllable
repeats). I then measure different aspects of the songs, how long are the
trills, what are their frequency ranges, how many trills and syllables are
in the song.
 |
| Example of a Lincoln's sparrow song showing the different parts (trills are outlined in red, individual syllables in purple). |
Like so many things in science, when I
first started to do these analyses I thought they were relatively routine,
clearly defined measurements. Then, as I encountered more songs, I began to get
ambiguous cases. Were the trills really trills or just long, complex syllables?
Was the song really a song, or just a complex call? One of the greatest sources of fascination in biology can
also be a biologist’s greatest source of frustration. Biology is all about
variation, and often the more you try to classify that variation the more
likely you are to find someone breaking the rules.
In my most recent project, I am trying to classify each syllable of a
song into a unique syllable type. This is both a fun and frustrating process
for Lincoln’s sparrow songs. So far, I’ve been classifying syllables by looking at spectrograms and listening to the songs.
 |
| Spectrograms of three different songs from different males. In each song, most of the syllables are unique, but all of the songs share one syllable type (outlined in red). |
I’ve got some
pre-established rules to classify syllables. Still, I come across cases
where it is difficult to decide whether two syllables are the same type or are different.
In these cases, I do what anyone does when they are faced with difficulty and
uncertainty: I get a second opinion. For now that second opinion is in the form of a human. In the future though, I hope that second opinion
will be in the form of a computer program that can objectively categorize
syllable types. That's right, I’m going to ask the machine.
If you’ve made it this far in the post (I’m
talking to you, Mom), you may be wondering, who cares about syllable types?
Well, I care about them for a couple of reasons. Once I can classify
syllable types, I can figure out if males sing different syllable types in
different contexts. For example, I can ask if males vary their syllable types
based on the presence or absence of a rival. Second, I can figure
out how males learn songs. Since juvenile males learn their songs from
adults, I can figure out what types of syllables juveniles choose to copy from adults by comparing their spectrograms.
*Want to learn more about spectrograms? Check out Bird Song Hero
