There has long been an association between music and birdsong; the twentieth century French composer Olivier Messiaen is well known for his use of carefully observed bird calls in his music, as explored in this short film by the London Philharmonic. His first such work was the composition for flute and piano Le Merle Noir (The Blackbird) [fig.1] written in 1952.
Numerous other composers have used transcriptions of the melody of different species in their compositions; five such pieces, plus Le Merle Noir, were listed in an article on the BBC classical music website. This is however a time-consuming process; Messiaen used recordings slowed down so that he could then turn the songs into notes by hand. With my own burgeoning interest in the song of a particular species, I have investigated the possibility of matching large archives of digital recordings with the latest sound analysis software to turn bird song into a format that could be notated automatically.
It is worth taking a moment to consider the point of bird song and why its study may be of more value than mere idle curiosity. There are two conflicting drivers at play in the development of the songs of an individual bird. Firstly, acquiring a series of calls and songs that are unique to that species, or conspecific. This is achieved partly through mimicry and partly through innate structures both physiological and neurological [Beecher]. This leads naturally to sufficient homogeneity that a non-avian listener, such as me, can also identify types of birds by their song. Secondly, developing an individuality of performance so that a female might chose the singer over another, less talented, male. This creates interesting variety in the songs both as differing versions by the same singer as well as differentiation between individuals, especially when establishing territory [Hyman]. This variation can sometimes include portions of song from different species, or heterospecific. There is a wide variation in structure across species from the sole use of closely-matched conspecific song through to refrains featuring improvisations upon both conspecific and heterospecific elements. This is an active area of study, especially with reference to the neurological development of an individual’s song repertoire and implications this may have for the acquisition of language in humans, which seems to follow very similar processes [Jarvis].
Over the past few years I have been lucky enough to see and hear many of these birds singing whilst I was at work or at home so have been able to correlate a specific song type with this species. This song is the main mating call performed almost exclusively by male birds before and during the mating season.
The usual form of this song is a sequence of alternating notes, often a jump of a third(as in While shepherds watched…) or a fifth(Twinkle, twinkle…). In the course of my study it has become evident that this structure is tightly conspecific across a wide geographical range. To help identify different versions of the song I have imagined it being performed to repetitions of the word “thirty-two”, starting and finishing either on the lower note or the upper note. The four versions of the simplest song can then be notated. [fig 3]
By listening over several seasons I gained the impression that one could characterise the main mating call of the Great Tit with four characteristics: pitches used, whether the start or finish are on the upper or lower note, the average length of refrain, and any irregularities from the simplest form. Here is a short recording of a bird from Sweden singing a clear example of the simplest form and a musical transcription [fig.4] of his song:
He clearly sings on the notes Ab and C, always starting on the lower note and predominantly finishing on the upper note, mostly 7 or 8 note bursts with no irregularities.
This next one from Finland [fig.5] sings fifths on F# and C# in the inverted form that always starts at the higher pitch and mostly finishes there, in phrases of around 10 notes with no irregularities.
Now we introduce a significant variation that I have heard both in the wild and in numerous recordings. One of the warning calls of the species is a quickly repeated single note often followed by a long chirr as in this example from Germany.
It seems that some birds incorporate this into their song which can be represented in our notation as a pair of quavers. Here is a good version of this, recorded in Yorkshire. [fig.6] The analysis of this was a little harder and to hear the notes clearly I needed to slow the audio down, of which more later. However, he is singing in fourths (Hark the Herald…) on E and B, mostly starting on the upper and finishing on the lower, bursts of roughly eight sets of top and bottom, with the lower note repeated twice each time.
The final example is unusual but the bird was observed and identified by the recorder as a Great Tit. It seems to incorporate all the elements from the songs above and exemplifies well the use of notation to record his performance. I have cut the original recording to keep the clip relatively brief but he continued in a very similar vein for another half a dozen phrases. Despite the increased complexity, we can still use the four-part analysis: he sings Ab, C and Eb in thirds and fifths, starts and ends on the lowest, sings at least fifteen notes, starts in fifths, has a pair of thirds of C to Eb in the middle then ends in fifths again. However, this is starting to be limited in its scope and the notated score [fig.7] gives all that detail and more.
Up to now I have been sourcing birdsong by downloading files from Xeno-canto, which is a vast web-based repository of recordings catalogued by species. Here you can also view each sound file as a frequency spread or spectrogram (which is called a sonogram on the website) so that you can see the changes of note. [fig.8]
Each file was then put into a Java application called Tarsos which allows you to see the output over time as a waveform making counting notes and separating individual sounds easier [fig.9]. This program also lets you slow the sounds down without losing pitch which is vital with some of the shorter notes [Six]. I then manually transcribed the song into the score writing program Sibelius and annotated it into the versions you see above.
There are two main problems with this approach. Firstly, the Xeno-canto collection is naturally skewed towards more unusual bird songs as one will always want to post ‘interesting’ sounds rather than ‘normal’ ones much the same as with online photographic repositories. Secondly, the transcribing process is very cumbersome and error-prone. However, Tarsos has a midi output option which should convert the sounds into a music software friendly format. These can could then be transferred them into Sibelius and thence into notation; however I should at that point stop putting the ‘lyrics’ in! However, I have yet to get this to work and will seek assistance from the programmers. I also need to access banks of data that have been gathered from field studies where the birds are individually catalogued and recorded. This could allow, at a later date, song structure to be correlated with mating success and other metrics.
I believe I have demonstrated that musical analysis of bird song has genuine potential especially in a species like the Great Tit that performs in a manner that is largely compatible with western musical notation. Also, it is clear that there is the opportunity to automate some or all of the transcription process. Once that has been done I would hope that there will be other possible analyses available that could advance our understanding of the development, function and effect of birdsong.
Beecher, Michael: “Birdsong and Vocal Learning during Development“, Encyclopedia of Behavioral Neuroscience, volume 1, 2010, pp. 164–168, Elsevier
Johannes, Buhl: “XC117010 Great Tit“, Xeno-canto, http://www.xeno-canto.org/117010, Siebenbrunn, Germany, 7 Jan 2013, retrieved 28 Jan 2016
Franks, Rebecca: “Six of the best: pieces inspired by birdsong“, classical-music.com, http://www.classical-music.com/article/six-best-birdsong-pieces, 2 May 2014, BBC, retrieved 28 Jan 2016
Hyman, Jeremy: “Countersinging as a signal of aggression in a territorial songbird“, Animal Behaviour Volume 65, Issue 6, June 2003, pp. 1179–1185, Elsevier
Jarvis, Erich: “Learned Birdsong and the Neurobiology of Human Language“, Annals of the New York Academy of Sciences, Volume 1016, June 2004, pp. 749–777, Wiley-Blackwell
Litsgård, Mikael: “XC294695 Great Tit“, Xeno-canto, http://www.xeno-canto.org/294695, Gotland, Sweden, 25 May 2015, retrieved 28 Jan 2016
M***, David: “XC213266 Great Tit“, Xeno-canto, http://www.xeno-canto.org/213266, Helperthorpe, England, 10 Feb 2015, retrieved 28 Jan 2016
Matusiak, Jarek: “XC122905 Great Tit“, Xeno-canto, http://www.xeno-canto.org/122905, Warsaw, Poland, 26 Feb 2013, retrieved 28 Jan 2016
Messiaen, Olivier: “Le Merle Noir“, Score for Flute and Piano, 1952, Alphonse Leduc
musicanth: “Olivier Messiaen – Le Merle noir (The Blackbird) [Kenneth Smith, Matthew Schellhorn]“, YouTube LLC, https://www.youtube.com/watch?v=IhEHsGrRfyY, retrieved 28 Jan 2016
Paljakka, Eetu: “XC268100 Great Tit“, Xeno-canto, http://www.xeno-canto.org/268100, Uusimaa, Finland, 29 March 2014, retrieved 28 Jan 2016
Philharmonia Orchestra (London,UK): “Messiaen’s Use of Birdsong “, YouTube LLC, https://www.youtube.com/watch?v=0MgLXeaf3zc, retrieved 28 Jan 2016
Six, Joren et al: “TarsosDSP, a Real-Time Audio Processing Framework in Java“, 53rd AES International Conference on Semantic Audio, London, 27-29 Jan 2014, Audio Engineering Society
Wikipedia contributors: “MIDI“, Wikipedia, The Free Encyclopedia, https://en.wikipedia.org/wiki/MIDI, retrieved 28 Jan 2016
Wikipedia contributors: “Olivier Messiaen“, Wikipedia, The Free Encyclopedia, https://en.wikipedia.org/wiki/Olivier_Messiaen, retrieved 28 Jan 2016
Wikipedia contributors: “Spectrogram“, Wikipedia, The Free Encyclopedia, https://en.wikipedia.org/wiki/Spectrogram, retrieved 28 Jan 2016