Tag Archives: bird vocalization

Songbirds work around computational complexity by learning song vocabulary independently of sequence


(Dina Lipkind, Anja T. Zai, Alexander Hanuschkin, Gary F. Marcus, Ofer Tchernichovski & Richard H. Hahnloser 1 Nov 2017)


Songbirds, being skilled vocal learners18,19,20,21, provide an opportunity for studying how errors are assigned and minimized during the learning of complex motor sequences. A young zebra finch (Taeniopygia guttata) imitating an adult tutor has to match a series of spectrally distinct sounds (syllables) performed in a precise order (Fig. 1b). Zebra finches are capable of adjusting their developing song towards its target in a variety of ways, including morphing the spectral (phonological) structure of song syllables22,23,24,25, generating and adding novel syllables to their song23, 25, 26, and rearranging the positions of existing syllables26, 27. How then do they cope with the complexity of selecting the appropriate combination of operations that would reduce the mismatch between their own song and the target?

A possible way to reduce computational complexity could be to optimize one aspect of the task, while ignoring the costs of the other. At one extreme, the task could be reduced to assigning each syllable in the bird’s song to the temporally corresponding syllable in the target song (Fig. 1c, left). Such strategy would minimize sequence rearrangements, at the cost of possibly large phonological adjustments. Although this hypothesis has not been directly tested, a number of previous findings suggest that songbirds may not be using global alignment between song and target as a learning strategy. These include the observation that individual syllables are recognizable in developing zebra finch song before the correct sequence is apparent28; the existence of an early developmental phase in which repetitions of a single “proto-syllable” differentiate towards multiple targets22, 24, 25, 29, 30; the fact that many songbird species perform variable syllable sequences as adults (e.g., nightingales, starlings and Bengalese finches); and the ability of zebra finches to match a target exclusively through syllable rearrangements, without changing phonology26. An alternative strategy, therefore, could be to assign song syllables to target syllables in a manner that minimizes phonological distances, while ignoring combinatorial distances (Fig. 1c, middle). Such phonological greediness would increase the number of ensuing sequence changes and thus the overall sequencing cost26. An intermediate strategy could be to seek a trade-off between minimizing structural and temporal errors, for example by independently matching parts of the song sequence (such as phonology in bigrams or trigrams27) to parts of the target sequence

Read more


Penguins’ Calls Are Influenced by Their Habitat

AUK-17-75 D Colombelli-Negrel

(AOS 1 nov 2017;Photo: D. Colombelli-Négrel)

Birds use vocalizations to attract mates, defend territories, and recognize fellow members of their species. But while we know a lot about how variations in vocalizations play out between populations of songbirds, it’s far less clear how this variation affects birds such as penguins in which calls are inherited. A new study from The Auk: Ornithological Advances examines differences in the calls of Little Penguins from four colonies in Australia—nocturnal birds for whom vocalizations are more important that visual signals—and finds that disparities in habitat, rather than geographic isolation or other factors, seem to be the key driver of variation in the sounds these birds use to communicate.

Diane Colombelli-Négrel and Rachel Smale of Australia’s Flinders University recorded calls from four Little Penguin populations across a small area of South Australia, one of which had previously been shown to have subtle genetic differences from the other three, and used playback experiments to test penguins’ ability to distinguish between calls from different colonies. They found that agonistic calls, which are used in pair displays and aggressive situations, varied among the four populations, and that the calls’ characteristics appeared to depend on small-scale differences in the habitat where the penguins lived. However, birds did not discriminate between calls originating from different colonies, which suggests that agonistic calls don’t seem to play a role in isolating the two different genetic groups.

Penguins breeding in open habitats produced lower-frequency calls than those breeding in habitats with denser vegetation—the opposite of the trend typically observed in songbirds. The authors speculate that agonistic calls may be subject to different selective pressures because they’re used in close encounters with other birds rather than to communicate across distances, and could also be influenced by variation in the noise level of wind and surf. “I was excited to find that calls were influenced by habitat, as this hasn’t been investigated much in seabirds and most of our knowledge in this area comes from studies on songbirds,” says Colombelli-Négrel. “This new research suggests that many factors influence call variation in birds, which also depends on the function of the calls. This study highlights that many questions remain and that studies need to investigate more than one factor in conjunction with the function of the calls to fully understand call variation in seabirds.”

“This work tells an interesting story of vocal diversification in Little Penguins, and gives insight into how individual and micro-scale variation effects behavior,” according to Stony Brook University’s Heather Lynch, an expert on penguin calls who was not involved in the study. “Non-vocal-learning birds are relatively understudied in terms of vocalizations, and it is great to see penguin vocalizations being studied in such a way.”

How songbirds learn a new song

ETH Zurich, ScienceDaily; Photo1nov 2017

For a songbird, learning a new song is akin to a child learning a new language. Zebra finches approach this challenge step by step, and even make a detour in the process — by taking song syllables that they already know and adapting them to the syllables that they have to learn. During this learning phase, the syllable sequence often gets mixed up. The birds then arrange the newly-learned syllables into the correct order in the next learning phase. Researchers led by Richard Hahnloser, a professor at the Institute of Neuroinformatics run by ETH Zurich and the University of Zurich, have reported these findings in the latest edition of the journal Nature Communications.

“The zebra finches have evolved the strategy of dividing a task as complex as learning a new song into easy-to-manage parts,” says Hahnloser. “This allows them to expand their repertoire with minimal effort.”

The scientists made this discovery in an experiment with young birds that were less than a month old at the start of the study. On a daily basis, the researchers broadcasted a song to the birds, which the birds then learned. After a month, the researchers changed the song and the birds tried to adapt their song to the new one. “In nature, birds instinctively adapt their songs to those of adult birds of the same species,” explains Hahnloser. The researchers recorded all vocalizations made by the birds and used a computer to evaluate them syllable by syllable.

Sound examples

The letters represent different syllables (each of a specific pitch): + and — for a positive and negative semitone change, ++ for a whole tone change.

Bird that has mastered song ABC, with the task of learning song AC++B. In a first step, the bird changes the pitch of syllable C and sings ABC++. Only in a second step does the bird arrange the syllables in the correct order AC++B.

Bird that learned the song ABCB+ in the first place, with the difficult task of learning a second song AB++CB-. The bird must learn to sing syllable B both a whole tone higher and a whole tone lower. To do this, it makes a detour via the song AB-CB++. This means that it first changes the syllables each by a semitone (two small changes) and then tries to arrange the syllables into the correct order — which actually not a single bird achieved in the experiment. The birds reached adulthood during the course of the experiment; adult birds no longer change their song.

Computer linguistics with similar methods

“Interestingly, the birds’ strategy closely resembles the best methods currently used in computer linguistics to compare documents,” says Hahnloser. These algorithms compare written documents by considering their words in their context but regardless of their exact order. By comparing billions of texts, these algorithms can estimate the similarity of two words in terms of a number. In this way, for example, they identify that the words “house” and “building” have almost the same meaning.

These computer programs can also search through millions of documents to identify which is most similar to a given text, by determining which document contains the vocabulary that can be most easily adapted into the vocabulary of the comparison text. “Today’s computer scientists therefore use the same strategy that songbirds evolved — the birds have probably been using it for millions of years,” says Hahnloser.

Hypotheses for studies on humans

It is still uncertain whether infants use a similar approach when they learn their first or second language. However, Hahnloser believes that there are numerous similarities between the way humans learn to speak and the way birds learn their songs. Songbird research has already shown striking parallels with speech development in young children. For example, earlier studies showed that young birds and young children practise every single syllable extensively through duplication (with children, for example, “baba” or “dodo”). These studies showed that both continue to do so when learning new syllables even after they already mastered vocalizations containing two different syllables (with children, for example, “rabbit”).

Hahnloser’s latest research into songbirds has led to the hypothesis that young children also use a minimalist approach when it comes to learning a foreign language; they learn new sounds (for example, the nasal vowels or rolling of the R in French) by minimally adapting the sounds they already know, at first without considering the context of these sounds in the foreign language. However, further studies are still required to ascertain whether this is in fact the case.

Muscle fibers alone can’t explain sex differences in bird song

(Physorg 14 June 2017)

Male birds tend to be better singers than females—but does the basis for this difference lie in the brain or in the syrinx, the bird equivalent of our larynx? The researchers behind a new study from The Auk: Ornithological Advances analyzed the muscle fibers in the syrinxes of male and female birds from a range of species and found, to their surprise, that the amount of “superfast” muscle wasn’t typically related to differences in vocal ability between the sexes.

Most muscle fibers are one of two types—fast, specialized for short, intense bursts of activity, or slow, specialized for endurance. However, some animals, including birds, have a third type called superfast muscle that can contract around 200 times per second. Ron Meyers of Weber State University and his colleagues hypothesized that superfast muscle fibers in the syrinx might explain the greater singing ability of male birds, but when they analyzed the syringeal muscles of male and female birds from a range of species, they found that the amount of superfast muscle fiber didn’t differ between the sexes in most species. Instead, their results suggest that the role of superfast muscle is more complicated than they expected and may be related to the entire range of vocalizations of a species rather than song alone. Even though females of some species don’t sing, their superfast muscle fibers appear likely to play a role in the calls they use for other types of communication.

The researchers collected syringeal tissue from a total of ten bird species, some wild-caught and some from a University of Utah aviary. All species had both fast muscle and superfast muscle fiber in their syrinxes, but there was a clear sex difference in fiber type composition in only two species studied, Bengalese Finches and Zebra Finches. Based on this, the researchers speculate that the need for superfast muscle may be related to the entire vocal repertoire of each sex, not just singing behavior. Calls made by Zebra Finch females don’t have acoustic features that would require rapid muscle control, but in other species females may produce calls that require the muscle control provided by superfast fibers even if they don’t sing.

“The data really surprised us,” says Meyers. “Based on our first species studied, starlings and Zebra Finches, we went into this thinking that superfast fibers were related to singing in males. Zebra Finch males sing and females don’t, and males have 85% of the syrinx muscles made up of superfast fibers. In starlings, both male and females sing, and they both had about a 65% make-up of superfast fibers. But as the number of species we looked at grew, we had to totally change our perception of the role of superfast fibers in singing and the role they actually play in vocalizing.”

“Most of the research investigating the mechanisms of bird song focuses on the brain. However, research has begun to suggest that peripheral structures like the syrinx influence song divergence, which of course is an important factor that contributes to avian biodiversity,” according to Wake Forest University’s Matthew Fuxjager, an expert on superfast muscle. “This study therefore provides an exciting starting point to address this issue from a physiological perspective, and it shows that muscle fiber content in the syrinx might not be a strong predictor of avian vocal diversity. But then what is? I would argue that we’re still working this out, and that this study will provide an intriguing framework from which more work in this area can be conducted.”

Genetic differences across species guide vocal learning in juvenile songbirds


(Physorg 12 June 2017)

Juvenile birds discriminate and selectively learn their own species’ songs even when primarily exposed to the songs of other species, but the underlying mechanism has remained unknown. A new study, by researchers at Uppsala University, shows that song discrimination arises due to genetic differences between species, rather than early learning or other mechanisms. The results are published in Nature Ecology & Evolution.

Songbirds are our primary animal model for studying the behavioral and neural basis of vocal learning and memory formation in general. The tremendous variety in the songs of birds delights ornithologists and fascinates evolutionary biologists as a marker of species diversity. Explaining how species differences in song are maintained is a challenge because birds typically learn their songs by imprinting on songs heard when they were juveniles. What prevents juveniles from imprinting on the songs from a wide-variety of other species in their environment? When exposed to a mixture of different songs from their own and other species, juvenile songbirds discriminate and selectively learn songs typical of their own species, which suggests a remarkable fine-tuning of sound perception during the earliest stages of development. Despite the importance of these findings for our understanding of the vocal learning process, the mechanism underlying early song discrimination has remained unknown.

A new study by researchers from Uppsala University in Sweden resolves this mystery by first demonstrating that juvenile pied and collared flycatchers from the wild discriminate their own species’ songs before they’ve left the nest. Nestling flycatchers as young as 10 days old look at the sound source and produce more begging calls during experimental playbacks of their own species’ songs than to playbacks of the other species’ songs, demonstrating that song discrimination develops incredibly early in these two species. Next, the researchers swapped developing eggs from the nests of each species so that they were raised completely by parents from the other species. These nestlings also discriminated in favor of their own species’ songs, despite having no experience with their own species, demonstrating that song discrimination is not a result of early learning. Finally, to definitively show that genetic differences between species underlie discrimination, the researchers showed that hybrid nestlings formed as a result of matings between parents from each species discriminate in favor of the songs of one of the species, the pied flycatcher. Taken together, these results show that song discrimination has a genetic basis.

‘Song differences across species are vital for birds to choose appropriate mates and negotiate complex social interactions. A genetic basis for song discrimination in early life may help explain how song differences are maintained in a noisy, diverse world’, says David Wheatcroft, researcher at the Department of Ecology and Genetics at Uppsala University and co-author of the study.

The song learning process in birds and the language learning process share remarkable behavioral and neural parallels. One of the longest standing problems has been to determine how the brain encodes the vocal memories that underlie learning. The results of this study suggest that this process begins with a genetic blueprint expressed early in life.

Birds change song to be heard above traffic noise

(Julia John 2 may 2017; Photo Kelly Colgan Azar )

Vehicles are a major source of noise pollution for urban wildlife. That’s particularly a problem for birds that have to compete with the roar of engines to communicate. Recent research from Washington, D.C., suggests that some birds —those with innate rather than learned songs —modify their song structure to be heard over the din of traffic.

In noisier conditions, these birdsongs are shorter and have a smaller range of frequency, said Katherine Gentry, lead author on the paper published in Bioacoustics. The birds raise their minimum frequencies, reducing their song’s overlaps with low frequency traffic noise.

“That makes it easier for the receiver to detect the signal against the background noise,” said Gentry, a research assistant at George Mason University.

Gentry wanted to see if suboscines, a family of birds whose song is innate rather than learned, could adjust their signal to communicate over traffic noise. Suboscine species include the vermilion flycatcher (Pyrocephalus rubinus), chocolate-vented tyrant (Neoxolmis rufiventris), white monjita (Xolmis irupero) and eastern wood pewee (Contopus virens). Although many studies have examined birds’ response to noise pollution from vehicles, this was the first to look at how suboscines in particular change their song in the absence of traffic during temporary road closures.

Gentry recorded the song of the eastern wood pewee in Rock Creek Park, a large urban park in Washington, D.C. She then compared recordings from times when traffic noise was high and times it was low, including weekends, when roads were closed to vehicles to allow for cycling and joggers. “This study showed that suboscines can adjust their song as traffic noise fluctuates,” Gentry said. That’s good for the birds in that it allows them to improve their chances of successful communication, she said.

“Even though they’re improving the likelihood of signal reception, they’re potentially sacrificing the sexiness of their song,” Gentry said. “With that minimum frequency raised, they’re not singing that naturally selected song structure.”

Songs with a higher minimum frequency that have a narrower range of frequencies could negatively influence the birds’ ability to defend territories, find mates and reproduce if females and rival males respond more strongly to signals with wider bandwidths.

But Gentry also found that when the roads were closed and it was quieter, the birds sang more naturally. Even though a permanent reduction in traffic noise is most ideal, she said, road closures could benefit birds like these that alter their songs in response to higher traffic noise levels.

“Noise is an issue for animals,” Gentry said. “Even if they cope through signal adjustment, it could be affecting them in the long run. It’s important we make every effort to reduce it. It would be a broad benefit for the community.”

Research teaches machines to decipher the dawn chorus

(Author: Physorg; Photo: Wikipedia; 20 March 2017)

Innovative research looking at the timing and sequence of bird calls could provide new insight into the social interaction that goes on between birds. It will also help teach machines to differentiate between man-made and natural sounds and to understand the world around them.

The work is being led by Dr Dan Stowell, a research fellow in machine listening at Queen Mary University of London (QMUL). It is supported through an Early Career Fellowship from the Engineering and Physical Sciences Research Council (EPSRC). An audio slide-show Deciphering the dawn chorus on this research is available on YouTube with examples of the bird recordings.

In August 2015, Dr Stowell’s technology was released to the public in a smartphone app called Warblr. Users record the sound of a bird on their mobile device and the app analyses the sound, matches it with patterns of bird calls in its dataset and provides a list of possible species that the recording matches.

Dr Stowell is now building on this work to take the computer analysis of the sounds that birds make to a new level, to discover more about what messages are being communicated and who is dominating the conversations that are going on.

“Traditionally you would take explicit measures of things like how long is this sound, what frequency is that sound?” says Dr Stowell. “To go beyond this we use modern machine-learning methods where you don’t necessarily know how a computer has made a decision about a particular sound, but by training it, which means showing it lots of previous examples, we can encourage a computer algorithm to generalise from those.”

At the University’s laboratory aviary, female zebra finches provide Dr Stowell with plenty of audio examples for his work.

“We have put the timing of the calls together with acoustic analysis of the content of the call, for instance whether it is a short or long call. We examine factors such as does the probability of one bird calling increase or decrease after another bird calls or is there a more subtle interaction going on? Working out how strongly each bird influences another helps us to build a picture of the communication network that is going on in that group of birds.”

Read more