Northern Cardinal Songs

  1. Overview

  2. Tanglewood songs

  3. Individual repertoire

  4. Mount Airy songs

  5. Tanglewood- Mt Airy transect

  6. Songs from other locations

  7. Change with time

  8. Singing by females

  9. Methods

  10. A HoFi learned NoCa songs


In Cincinnati, where I live, Cardinals start to sing in earnest each year around the end of February or beginning of March. Even if the weather is especially cold and the arrival of spring delayed by masses of arctic air the increasing length of the days triggers hormonal changes in the birds and a behavioral consequence of these changes is that they --- especially the males --- perch in conspicuous locations like tree tops and sing loud songs typical of the species. While the exact nature of the song changes from one bird to the next, all Cardinals seem to recognize all the songs given by their neighbors. Humans also recognize most cardinal songs.

Among the singing birds (passerines, including the Cardinal) songs are believed to serve two purposes: the yestablish territories and they attract mates. So the lengthening days and the onset of singing behavior marks the start of the Cardinal's breeding season. Within a matter of weeks or a month territories will be established, mates attracted, nests built, and eggs laid. Singing continues into July, well after the young of the year have left the nest. It is during this period that new Cardinals learn how to sing from hearing the songs of the adults around them.

It is this social trasnmission of song and its consequences that makes the singing of Cardinals so fascinating to me. A slow spatial diffusion of song types occurs due to learing of songs from neighbors. This is coupled with a more sporadic but potentially faster dispersal of songs associated with the movement of young birds away from their natal territory.

What is a song?

Songs are the Cardinal's most conspicuous vocalizations and are distinguished from other sounds, varoiusly referred to “notes”, “calls”, and “chips”, by their complex structure and the volume at which they are delivered. Songs serve the purpose of advertisement; they're used to declare and defend a breeding territory and to attract mates. In contrast, other notes seem to serve the purpose of more covert communications such as maintaining contact with flock mates, warning of the presence of preditors. Further, songs are delivered while perched, often in a consicuous locations, while other notes are often given from the ground, while hidden in foliage, and while feeding.

Here are two song types that were sung by the same bird during the same bout of singing. Two repetitions of each type are shown and they illustrate the fact that the variations among repetitions of a single type of song are minor compared to the differences between songs of different types.

These songs illustrate a feature common to all NoCa songs. Each is made by a a repetition of a particular element, followed by a repetition of another element. Usually there are just 1 2 or 3 different elements that occur in a song. We recognize these elements be cause they are the unit of repetition.

Below I'll described a fairly techincal procedure for deciding when two songs are the same and when one pair of songs is more similar to each other than another pair. But for most ractical purposes we can trust our ears and/or a glance at a spectrogram to decide the question.

Is it a single entity or combinatorial? Repetition of elements a different number of times (but in the context of singing the same song type) suggests that the particular number of repetitions isn't too important. What do NoCa recognize as songs? What do people recognize as NoCa songs? Synthetic songs Should we make an inventory of different songs or of different elements? Are there just some finite number of different songs, or is there continual innovation that would in principle eventually result in unlimited numbers of different songs? We might classify songs into types according to the number of different elements they contain Are there an infinite number of NoCa songs, or only a finite list?

Repertoire of an individual

Songs of a region. Tabglewood, Mt Airy

Songs through time

Singing by females

Is our recognition of songs objective?

(Learning in a noise-filled environment)

I've been listening carefully to the songs of Cardinals around the house for many years and recording them for almost as many. The ability record and then make spectorgraphs that illustrate bird songs helps tremendously in comparing and in learning to recognize the songs. Here my subject is the diversity and geographic and temporal distributions of different types of Cardinal songs. So you'll find many spectorgraphs that illustrate songs. I'f you're not accustomed to these displays, never fear, since each spectrograph is accompanied by a link that will let you actually hear the song.

Each Cardinal can sing a number of different songs. It isn't usually very hard to distinguish between two different songs. For example, here are two different songs sung within seconds of each other by a single bird sitting on a single perch:






The repertoire of a single bird runs to at least some 8 or 10 different songs. In any one location, most of the Cardinals sing similar songs; they share most of their repertoire with their neighbors.

Northern Cardinals learn their songs and each bird can sing a number of different songs. At any given location near where I live (in Cincinnati, Ohio) one can hear about 10 different types of song, at least if one's patient. And, the songs one hears changes from location to location.

When neighboring Cardinals are singing at the same time it is very likely that they will sing the same type of song; certainly much more likely than would be expected by chance if they chose from among all the songs they can sing at random. This could be in part because they learn songs from each other and so neightbors tend to know similar songs. Or it could be because they never move far from where they hatched and originally learned their songs. Or perhaps they actively choose to sing songs that nearby birds are singing.

Sometimes one can distinguish an individual by the way it sings a particular type of song; differences between the songs of different birds can be fairly large and the variation in the songs from an individual tend to rather small. Similarly, all the songs of a type sung at a given location tend to be more similar to each other than to songs of the same type sung at a remote location. And the types of songs sung at one location has only partial overlap with the songs sung at another site.

Here I've used sonograms and sound clips to illustrate different types of song. Traditionally, differences of this sort are established by making a few measurments (lowest frequency, highest frequency, number of syllable repetitions, etc.) on a number of different songs and then using a multivariate statistical procedure on this data. But basic human ability to recognize patterns is sufficient to distinguish most song types. The song types displayed here correspond well to those identified by the "yardstick" procedure and module that is provided by the program Dora2

Song types from Tanglewood Lane

Roughly speaking a song type is a collection of similar sounding songs. The songs of a type are understood to be more similar to other songs of the type than to any other songs. To be considered a song type, the song must come from several different birds. Ideally one should have evidence that the birds treat the songs as similar by using them at the same time when two or more birds a singing simultaneously. In practice it is pretty easy to classify songs to type by examining sonograms and using general notions of pattern recognition.

The vast majority (over 95%) of Cardinal songs from this location can be assigned to one of 10 different types. Here, songs from each of the 10 most commonly heard types are illustrated, each by several examples that are taken from different indivuals and/or different years.

Cardinal songs are typically made up of 2 different syllables, each repeated several to many times. Most syllables are actually a slurred note or notes, so you can spot different syllables by noting the pattern of rising and falling pitch. A few syllables are made up of a few notes on different pitches. In each song, note the number of different syllables, the number of times each is repeated, and the rapidity with which they are repeated.

  1. Type T1

    Type T1

    play 01-02.mp3

    play 2004-md-02-55.mp3

    play noca-05-md-02-59.mp3

  2. Type T2

    Type T2

    play 04-md-01-04b.mp3

    play md02-15.mp3

    play noca-05-md-01-109.mp3

  3. Type T3

    Type T3

    play 01-18a.mp3

    play noca-05-md-02-104a.mp3

    Type T3 variant

    play 01-18b.mp3

    play noca-05-md-02-02.mp3

  4. Type T4

    Type T4

    play 2004-md-02-91.mp3

    play 2004-md-02-92.mp3

    play noca-05-md-01-102.mp3

  5. Type T5

    Type T5

    play md01-18a.mp3

    play 04-md-01-06b.mp3

    play noca-05-md-02-88.mp3

  6. Type T6

    Type T6

    play 2004-md-01-65.mp3

    play md01-08.mp3

    play noca-05-md-02-101.mp3

  7. Type T7

    Type T7

    play 2004-md-01-88a.mp3

    play 01-05a.mp3

    play noca-05-md-01-51.mp3

  8. Type T8

    Type T8

    play md01-18b.mp3

    play 2004-md-02-52.mp3

    play noca-05-md-01-48b.mp3

  9. Type T11

Type T11

play 2004-md-02-54a.mp3

play 2004-md-02-57.mp3

play 2005-md-01-33(noca)B.mp3

Individual repertoire

The best evidence that a bird can sing several different types of songs is a recording in which this actually happens. As Cardinals will repeat a single type of song many times before switching to a different type, I have very few continuous recordings of a single bird singing more than two song types. Still, this sort of evidence supports the conclusion that most birds sing at least two kinds of song and a fair number of them sing three or more.

Cardinals defend nesting territories and sing from particular perches in their territory repeatedly throughout the day. They tend to use the same perch day after day as well. Recordings of a birds singing different songs from the same perch can also be thought of as evidence that a single bird sings different songs. Based on this assumption several Cardinals I've encountered can sing 5 or 6 different kinds of song.

Finally, some birds sing an unsual and very distinctive form of a particular song type --- a form so unusual that it is probably characteristic of that individual. This peculiar variant, when it occurs in different recordings together with a number of different different song types provides evidence that all these song types belong to a single bird's repertoire.

For example, here is a Type T2 song sung by a bird through much of February and March 2005. While clearly type T2 (and often sung at the same time that other Cardinals were singing type T2 songs) this version is distinctive because of the nature of its last few elements. Type T2 play noca-05-md-02-87.mp3

Below are songs that were sung at Tanglewood over the course of several days in the spring of 2005. They were all sung from the same perch as the distinctive song above and were all sung in conjunction (appear in a continuous recording from a single bird) with a song of this distinctive type.

Type T3 play noca-05-md-02-104a.mp3

Type T3 variant play noca-05-md-02-104c.mp3

Type T11 play noca-05-md-02-104d.mp3

Type T6 play noca-05-md-02-101.mp3

Type T5 play noca-05-md-02-88.mp3

In addition, I heard but did not manage to record, the bird sing a type 1 song.

Singing Types 1,2,3,3var,5,6,and 11 gives this bird a repertoire of at least 7 song types. This is the largest repertoire I've observed for a Northern Cardinal.

Mount Airy

Mount Airy Arboretum is far enough west of Tanglewood (about 1 mile) that no sane Cardinal would move from one location to the other. The density of Cardinals at the two locations is roughly the same; under optimal conditions one can hear 6 or 8 birds singing simultaneously.

These examples of 10 song types from Mt Airy Arboreturm illustrate the variation between songs at different locations. Some song types occur at both Tanglewood and Mt Airy (Type T2, T4, T6, T7, T11) while others are unique to either one or the other of the locations. Of the types that occur in both places, the differences between the locations seems (subjectively) to be larger than the differences among the songs of the type form Tanglewood.

Type 2 (MA2)

play noca-05-md-02-110aType MA2 This looks like a type T2 song

Type 4 (or MA4)

Similar to type T4 except 1st elements differ.

play noca-05-md-02-32

Type MA6

Most like a T6 song, especially the ending elements.

play noca-05-md-02-67

Type MA7

Most like a T7 song.

play noca-05-md-02-77

Type 11

play noca-05-md-02-15Type MA11 Like a T11 song

  1. Type MAV

    play 2004-md-01-31

    play noca-2003-md-02-06

    Type MAW

    Currently unknown from Tanglewood

    play noca-05-md-02-80

    Type (MAX)

    play noca-05-md-02-12 type MAX

    play noca-2004-md-01-33

    Type (MAY) =type 1=MA1

    Most similar to type T1 element 2 and ending element of T4 and and MA4. I've now (march 06) decided this is a type 1 song that is missing the starting elements.

    play noca-05-md-02-20type MAY

    Type (MAZ)

    Maybe a switch in song type in second song.

    play noca-05-md-02-119

    Type "8a" =MAX

    I've now decided (march 06) that this is just type MAX.

    play noca-2004-md-01-21 Type "8a" from Mt. Airy

The Tanglewood - Mt Airy transect

While Cardinals probably do not fly from Tanglewood to Mt Airy, there's a corridor of Cardinal habitat connecting the locations and so each NoCa is the neighbor of a neighbor (... of a neighbor..) of a NoCa living in the other location.

Some types of songs occur at both ends of the transect and typically the differences between songs of a type from the ends of the transect are greater than between two songs from one end or the other. It'll be interesting to see if there's a gradient connecting the two "subtypes" of songs. T7->MA7 and T2->MA2 are good examples. T4 and MA4 are very similar except for their 1st elements and these are quite different.

Songs that aren't part of a gradient running from T to MA must be restricted to geographic regions whose boundaries intersect the transect. Are these boundaries randomly distributed or do they accumulate at specific locations (like at the places where the transect crosses a busy road?

Northern Cardinal songs from other locations

There are several reasons to suspect that there's something "universal" about NoCa songs. Perhaps the most obvious is that humans reconize them fairly easily, even if they are of a type never heard before. Another is that the same types seem to occur all over the place, but the region in which a type occurs is not connected. Here I've just illustrated NoCa songs from other locations and sometimes provided an annotation indicating to which Tanglewood-Mt Airy transect I consider it related.

An interesting experiment would be to generate synthetic NoCa songs using real song elements and determine whether humans can identify them. As a hypothesis: songs that are generated in the form AABBBBB where A is a starting element of a real NoCa song and B a 2nd element would be found, by humans, to be NoCa songs, but BBAAAAA would not. Of course, it'd also be worth trying it with playback on real birds as well.

  1. Miami Whitewater Forest, Hamilton County, Ohio

  2. Cincinnati Nature Center, Clermont County, Ohio

  3. Naples, FL

  4. Kearney NE

Man-made noise

If NoCa are learning their songs in a noisey envirnoment, some acoutic or psychological effect, interference or masking of example, could make them more liekly to learn some types rather than others. Or to learn altered or modified versions of songs. Is there evidence that the songs sung in quiet and noisey enviroments are different?

Drift with time

Here are a few recordings from Tanglewood during the period 1999-2001. They indicate that several of the song types persisted at that location for at least 5 or 6 years. I imagine that the life expectancy of a NoCa is probably considerably less than 5 or 6 years, but it is possible that these songs are actually from birds that were still singing in 2005.

play 99_1_08(noca-T2)Compare to type T2 songs.

2000_02_12(noca-T4)Compare to type T4 songs.

2000_02_23(noca-T5)Compare to type T5 songs.

2001dat1-38A(noca-T7)Compare to type T7 songs.

Female NoCa sing frequently. Lower perches, more variable song elements, greater tendency to switch song types in the middle of a song.



Recordings were made using an omnidirectional microphone in a plastic, 24 inch parabolic reflector. All the recordings from Tanglewood were made within about 50 yards of our front door; all the recordings from Mt Airy were made within about 300 yards of the parking lot at the Arboretum. In 2003 I used a cassette tape recorder, in 2004 and 2005 I used a Minidisk recorder (and, occasionally, a dat recorded). You'll notice the difference in background noise levels in the recordings made with the analog and digital devices. Recordings were transfered to the computer using "line out" on the cassette recorder and "headphones out " on the Minidisk. I used the computer's sound card to digitize the recordings; it was set to take 16 bit samples at 44.1 KHz. Sonograms were generated from the digitized sound files using the Dora2 program.

Song probes

Song probes are an approach to making an objective definition of "song type" and thus for making automatic identification of song types possible. The idea is to take small bits (elements) of real songs and use them as probes by taking an entire song and computing the correlation (or similarity) of the song with each of the probes. Then a multivariate analysis is possible and some clustering or pattern recognition procedure can be used to identify the type of the song.

The "Library" module of Dora2, also available as a stand-alone program called Library computes correlations between sonograms of different sound files. The sonograms of two sounds are generated exactly as in the Sonogram program, including the final scaling that reduces the magnitude of each Fourier coefficient to a number in the range [0,1]. This produces two arrays of numbers. one for each sound. The Library program computes the correlation coefficient between the numbers in the two arrays as well as the correlation between all shifts of one array relative to the other assuming only that they contune to overlap completely. That is, the array from the shorter sound is shifted against the longer, with r computed for each shift.

It reports the maximum of these r's, and these can be used as a measure of the similarity of the sounds.

I generated a collection of probes for songs from Tanglewood by selecting one example of each of the types that I recordered there in 2005. Each sound was split into separate phrases (elements) and one element of eact type was selected as a probe. In addition, for elements that are repeated many times, I also created probes that contained two consecutive elements from a repetion. (This provides some measure of the timing or rate at which the elements are repeated). I also created a second set of probes based on 1 recording of each type from Mt Airy in February 2006.

The naming scheme for the probes is, roughly >i>VIa-1 which means that the probe is from as type VI (6) song, that it came from the first (a vs. b vs. c etc) example of such a song, and was an element of the 1 st type. Probes based on songs from Mt Airy have nanes starting with "M".

As a test of these probes, I took samples for the recordings of NoCa from Tanglewood and Mount Airy in 2006 from the month of Februay, selected one full song from each, and used the Library program to find the correlations (similarities) of each of these songs with each of the 2005 Tanglewood probes. The results is an array with 109 rows (one for each 2006 song recorded before 11 March at either Mt Airy, Tanglewood, or on the transect between them) and 69 columns, one for each probe. Using the the values 1-r and a euclidean measure of distance, I processed the array using the hclust routine from R. This routine performs complete hierarchical clustering in an attempt to group rows that are most similar. The results are shown in the figure:

The tips of the resulting tree are labeled with the type I identified for each recording. Tte prefixes "t","m","tr",and "b" are applied to songs from different locations: Tanglewood, Mt Airy, Tanglewood Ridge, and the powerline cut Between Colerain and Kirby Roads. In addition, several songs that appeared to be of two types were labeled "xn/m" where "n" and "m" are the two types represented.

I consider this clustering pretty good performance. With the exception of tentatively defined "new" song types like MAS,MAZ, MAX from Mt Airy, songs of each "type" I've identified cluster together. This suggests that automatic type identification is possible. The performance of this method could certainly be improved by restricting analysis to songs with only low background noise.

  1. For the most part, all the main types of songs cluster together. Where they don't, I've perhaps made a mistake in assigning a song type.

  2. As would be expected the compound "xn/m" songs cluster either with type "n" or type "m" songs.

  3. Type MAR clusters together inside the type 2 cluster and closest to a type m2 song. This is based partly on the similarity of the 1st elements, and the steep vertical slurs in the 2nd elements. Perhaps MAR is derived from type 2? I recall finding a NoCa at Tangelwood that had no little low frequency hooks on it's type 2-2nd element notes.

  4. One type M7 clusters external to a group cointaining all the other type 7 and type 1 songs. This is an unusual recording with lots of noise and echo as well as conspicuous overtowns due to near saturation. It should probabely be excluded from analysis based on the poor quality of the recording.

  5. Two MAS and one MAZ cluster with type 8. The MAS,MAX, MAZ complex is still tentatively defined, poorly resolved, and perhaps needs reclassification. Could MAZ and MAS be the same? Look at some older MAZ samples or get more new ones.

    Songs a given type (eg types 4,6,7, and 1) are clustered together according to locations, suggesting that there is some significant differentiation between the same type of songs from different locations. Type b7 clusteres with type t6 songs and is quite different from other type 7 songs.

Measuring sound levels

I don't have equipment to measure absolute sound levels, so I rely on making controled sound recordings (always using the same recorder, record level (preamp) setting, microphones, microphone mounts, and wind screens) and digitizing them in exactly the same way (playing the recordings back on the recorder with the same settings, through the same cables, and digitizing the sound using the same soundcard and program settings.) I convinced myself that this procedure works pretty well by generating 1 minute of white noise and then making several versions of this recording with (digital) amplifier settings of 0,-5,-15, and -25 dB. Indoors, in a fairly quiet room, I played each of these recordings back 3 times using the same amplification and recorded the playbacks after readjusting the recorder settings. Each of these recordings I then digitized twice according to my standard procedure. Then I used the SoundMeter program (BlockSize=1024) to measure the log of the average size of the samples in the recordings. A linear mixed effects model (in R using package nlme) showed that the association between the original loudness of the recordings (dB) and the logs of the mean sample sizes was very strong and linear. The variances attributed to replication of the recording and digitization procedures were extremely small. A slight departure from a linear relationship could be seen in the quieter recordings and this was consistent with the explanation that other noise in the room (the computer's fan, for example) were nearly of the same magnitude as the quietest recording.

You don't want all the details. The main conclusion is that the recording and digitization procedure is very reproducable and the scheme implemented in the SoundMeter program works well to generate measurements that reflect the actually sound levels at the time of the recording.

The SoundMeter program reads an audio file (in the .wav format) in blocks of BlockSize (typically 1024, but user-adjustable) samples and finds the mean square of the samples in the block. It then takes the log of this average. The program reports the min, max, mean, and standard deviation of the log's of the means of the squares of the samples.

Open questions

... that I'm currently interested in.

  1. Are different types of song more likely to be used at different times of the day or year?

  2. Does man-made noise (air planes, traffic) have an effect on the singing behavior of Cardinals? First, are there differences between the overall singing behavior (diversity of songs, types of songs, etc.) that NoCa use in locations with lots and little noise? Might we expect that sounds better adapted to the environment would predominate so that we'd find shifts in sound frequencies used or alteration in the frequency with which different song types are employed? Second, is an individual bird's singing influenced by other sounds? Is a bout more likely to stop if a loud sound source is present?

  3. Is the number of song types found at a location related to the density of Cardinals there?