Automated birdsong recognition in complex acoustic environments: a review

Conservationists are increasingly using autonomous acoustic recorders to determine the presence/absence and the abundance of bird species. Unlike humans, these recorders can be left in the field for extensive periods of time in any habitat. Although data acquisition is automated, manual processing of recordings is labour intensive, tedious, and prone to bias due to observer variations. Hence automated birdsong recognition is an efficient alternative. However, only few ecologists and conservationists utilise the existing birdsong recognisers to process unattended field recordings because the software calibration time is exceptionally high and requires considerable knowledge in signal processing and underlying systems, making the tools less user‐friendly. Even allowing for these difficulties, getting accurate results is exceedingly hard. In this review we examine the state‐of‐the‐art, summarising and discussing the methods currently available for each of the essential parts of a birdsong recogniser, and also available software. The key reasons behind poor automated recognition are that field recordings are very noisy, calls from birds that are a long way from the recorder can be faint or corrupted, and there are overlapping calls from many different birds. In addition, there can be large numbers of different species calling in one recording, and therefore the method has to scale to large numbers of species, or at least avoid misclassifying another species as one of particular interest. We found that these areas of importance, particularly the question of noise reduction, are amongst the least researched. In cases where accurate recognition of individual species is essential, such as in conservation work, we suggest that specialised (species‐specific) methods of passive acoustic monitoring are required. We also believe that it is important that comparable measures, and datasets, are used to enable methods to be compared.     

Birdsong Denoising Using Wavelets

Automatic recording of birdsong is becoming the preferred way to monitor and quantify bird populations worldwide. Programmable recorders allow recordings to be obtained at all times of day and year for extended periods of time. Consequently, there is a critical need for robust automated birdsong recognition. One prominent obstacle to achieving this is low signal to noise ratio in unattended recordings. Field recordings are often very noisy: birdsong is only one component in a recording, which also includes noise from the environment (such as wind and rain), other animals (including insects), and human-related activities, as well as noise from the recorder itself. We describe a method of denoising using a combination of the wavelet packet decomposition and band-pass or low-pass filtering, and present experiments that demonstrate an order of magnitude improvement in noise reduction over natural noisy bird recordings.     

HABITAT ACOUSTICS OF A NEOTROPICAL LOWLAND RAINFOREST

ABSTRACT

The acoustic characteristics of an Amazonian lowland rain forest study site in southern Venezuela was analysed to determine environmental constraints upon acoustic communication. Signal degradation was measured by conducting transmission experiments at different heights above ground level. Measurements of ambient noise served to determine possible communication distances for various times of day, heights above ground level and frequencies. “Sound windows” for acoustic long-range communication were found for low frequencies, calling heights in the midstorey and calling in the morning or during the night. Sound attenuation was affected by height and frequency but not by time of day. Background noise varied remarkably with time of day and frequency and had a greater impact on communication distance than signal attenuation.     

Body size correlates negatively with the frequency of distress calls and songs of Neotropical birds

ABSTRACT The allometric relationship between body size and song frequency has been established in previous studies of temperate and tropical bird communities. However, the relationship between body size and the frequency of distress calls has been examined in only one study of temperate birds. We examined size‐frequency relationships in the distress calls and songs of a Neotropical bird community in northwestern Costa Rica. In 2008 and 2009, we recorded distress calls and determined the body mass of 54 mist‐netted birds representing 38 species, 35 genera, and 14 families. We obtained songs for these same species from sound libraries and commercially available compact discs. For each vocalization, we measured minimum frequency and frequency of maximum amplitude. Larger birds produced lower‐frequency distress calls and songs than smaller birds. Phylogenetically controlled analyses revealed that the frequency of maximum amplitude was negatively correlated with body mass for both distress calls and songs. Minimum frequency was negatively correlated with mass for distress calls, but not songs. Our analyses suggest that the influence of phylogeny on the relationship between frequency characteristics and body size is modest. Pair‐wise comparisons across 37 species revealed that distress calls and songs had similar minimum frequencies, but songs had significantly lower frequencies of maximum amplitude than distress calls. This difference may arise from differences in signal function. Lower‐frequency sounds should transmit farther through forest habitats and songs must often transmit longer distances to reach their intended audience than distress calls. Our results support the general theory that body size is negatively correlated with the frequency of acoustic signals by demonstrating that this pattern holds true for both distress calls and songs in a Neotropical bird community.     

Neural function of the mesencephalic dorsomedial nucleus (DM) on distance call production in Bengalese finches

In sexually mature Bengalese finches, acoustic structures of distance calls show sexual difference. The dorsomedial nucleus (DM) of intercollicular complex is known as the midbrain vocal center of distance calls. Neural input from the robust nucleus of archistriatum (RA) was observed in the DM of sexually mature males, but not observed in that of sexually mature females. The purpose of this study is to clarify somo more details of physiological function of the neural system in the DM in distance call production. Electrical stimulation to the DM of both sexes induced a call acoustically similar to distance calls, whose duration depended on the number of the pulses/train of electrical stimulation; electrical stimulation in relatively large (or small) numbers of pulses/train induced calls with relatively long (or short) duration, respectively. Multi-unit spikes were recorded from neurons in the DM. The increment of the frequency of recorded spikos was large when the bird vocalized distance calls, and the number of the frequency decreased when the bird vocalized calls whose duration was shorter than that of distance calls. These results suggest that the neural system in the DM controls duration of distance calls in sexually mature males and females. Electrical stimulation to the DM under different pulse frequencies induced calls with different patterns of time-frequency characteristics. The relation between the pulse frequencies and time-frequency characteristics showed sexual difference. The relation between them in RA-lesioned males was similar to that in females. These results suggest that the neural circuit in the DM of sexually mature males is consisted of sexually common neural circuit controlled by the neural input from the RA, and that these sexually different neural system produce sexually different acoustic structures of distance calls.     

Weather conditions determine attenuation and speed of sound: Environmental limitations for monitoring and analyzing bat echolocation

Echolocating bats are regularly studied to investigate auditory‐guided behaviors and as important bioindicators. Bioacoustic monitoring methods based on echolocation calls are increasingly used for risk assessment and to ultimately inform conservation strategies for bats. As echolocation calls transmit through the air at the speed of sound, they undergo changes due to atmospheric and geometric attenuation. Both the speed of sound and atmospheric attenuation, however, are variable and determined by weather conditions, particularly temperature and relative humidity. Changing weather conditions thus cause variation in analyzed call parameters, limiting our ability to detect, and correctly analyze bat calls. Here, I use real‐world weather data to exemplify the effect of varying weather conditions on the acoustic properties of air. I then present atmospheric attenuation and speed of sound for the global range of weather conditions and bat call frequencies to show their relative effects. Atmospheric attenuation is a nonlinear function of call frequency, temperature, relative humidity, and atmospheric pressure. While atmospheric attenuation is strongly positively correlated with call frequency, it is also significantly influenced by temperature and relative humidity in a complex nonlinear fashion. Variable weather conditions thus result in variable and unknown effects on the recorded call, affecting estimates of call frequency and intensity, particularly for high frequencies. Weather‐induced variation in speed of sound reaches up to about ±3%, but is generally much smaller and only relevant for acoustic localization methods of bats. The frequency‐ and weather‐dependent variation in atmospheric attenuation has a threefold effect on bioacoustic monitoring of bats: It limits our capability (1) to monitor bats equally across time, space, and species, (2) to correctly measure frequency parameters of bat echolocation calls, particularly for high frequencies, and (3) to correctly identify bat species in species‐rich assemblies or for sympatric species with similar call designs.     

At sea vocal repertoire of a foraging seabird

Seabirds spend most of their time at sea, yet our knowledge of their activities and behaviour is limited due to difficulties of in‐situ data collection. In particular, we know virtually nothing about their acoustic communication when at sea. We benefited from the recent development of miniaturised audio‐recording devices to deploy acoustic recorders on breeding Cape gannets Morus capensis to study their vocal activity while foraging. Call sequences were recorded on 1718 occasions, from which acoustic variables were measured on calls with good recording quality. A total of 1348 calls from 18 birds were measured in temporal and frequency domains. Each call was assigned to a behavioural context defined acoustically: sitting on the water, flying, taking off or just before diving. Potential discrimination among calls from different contexts was tested using the random forest algorithm. Within each context, individual stereotypy in the calls was assessed per acoustic variable using a measure of potential of individual coding, and as a combination of variables using a similar multivariate analysis. The acoustic structure differed according to the behavioural context (global accuracy of prediction 75%). Temporal variables (sequence and call duration) were most important to correctly classify the calls among the four contexts. When considering only two contexts, on the water and in the air (merging flying and diving), frequency and spectral variables (percentage of energy below 1200 Hz and fundamental frequency) were of most importance (accuracy 86%). A combination of acoustic variables was necessary to discriminate individuals, but calls from all contexts were not strongly individually distinct (accuracy 41–63%). We provided the first detailed acoustic analysis of a foraging seabird and demonstrated context‐specific acoustic structure in its vocalisations at sea. Our results suggest that seabirds use vocal communication to exchange various types of information that likely improves foraging success.     

The effect of body size and habitat on the evolution of alarm vocalizations in rodents

When confronted with a predator, many mammalian species emit vocalizations known as alarm calls. Vocal structure variation results from the interactive effects of different selective pressures and constraints affecting their production, transmission, and detection. Body size is an important morphological constraint influencing the lowest frequencies that an organism can produce. The acoustic environment influences signal degradation; low frequencies should be favoured in dense forests compared to more open habitats (i.e. the ‘acoustic adaptation hypothesis’). Such hypotheses have been mainly examined in birds, whereas the proximate and ultimate factors affecting vocalizations in nonprimate mammals have received less attention. In the present study, we investigated the relationships between the frequency of alarm calls, body mass, and habitat in 65 species of rodents. Although we found the expected negative relationship between call frequency and body mass, we found no significant differences in acoustic characteristics between closed and open‐habitat species. The results of the present study show that the acoustic frequencies of alarm calls can provide reliable information about the size of a sender in this taxonomic group, although they generally do not support the acoustic adaptation hypothesis.     

Experimental test of birdcall detection by autonomous recorder units and by human observers using broadcast

相似文献   

The role of ecological constraint in driving the evolution of avian song frequency across a latitudinal gradient

Just as features of the physical and biotic environment constrain evolution of ecological and morphological traits, they may also affect evolution of communication systems. Here we analyze constraints on rates of vocal evolution, using a large dataset of New World avian sister taxa. We show that species breeding in tropical forests sing at generally lower frequencies and across narrower bandwidths than species breeding in open habitats, or at high latitudes. We attribute these restrictions on birdsong frequency to the presence of high-frequency insect noise and greater degradation of high-frequency sounds in tropical forests. We fit Ornstein-Uhlenbeck models to show that recent evolution of song frequency has been more greatly constrained in tropical forests than elsewhere, that is, songs have shown less tendency to diverge over time in tropical forests, consistent with inferred acoustic restrictions. In addition, we find that song frequency has evolved more rapidly overall at high latitudes in both forest and open habitats. Besides a larger available sound window, other factors contributing to more rapid divergence at high latitudes may include an overall increased intensity of sexual selection, occupation of more divergent habitats, and the presence of fewer competing species.     

The evolution of nonhuman primate loud calls: Acoustic adaptation for long-distance transmission

Many nonhuman primates produce species-typical loud calls used to communicate between and within groups over long distances. Given their observed spacing functions, primate loud calls are likely to show acoustic adaptations to increase their propagation over distance. Here we evaluate the hypothesis that primates emit loud calls at relatively low sound frequencies to minimize their attenuation. We tested this hypothesis within and between species. First, we compared the frequencies of loud calls produced by each species with those of other calls from their vocal repertoires. Second, we investigated the relationship between loud call frequency and home range size across a sample of primate species. Comparisons indicated that primates produce loud calls at lower frequencies than other calls within their vocal repertoires. In addition, a significant negative relationship exists between loud call frequency and home range size among species. The relationship between call frequency and range size holds after controlling for the potentially confounding effects of body size and phylogeny. These results are consistent with the hypothesis that nonhuman primates produce loud calls at relatively low frequencies to facilitate their transmission over long distances.     

基于语谱图特征信息分割提取的声景观中鸟类生物多样性分析

声学手段是监测和研究生态系统生物活动规律、评价生态系统健康状况的一种新方法,声景观生态学也是景观生态学的一个新兴研究领域。声景指数是描述复杂的音频数据生态学特征的有效方法,但是,单一的声景指数并不能有效的指示物种的真实丰度。在充分挖掘音频文件时频结构特征的基础上,将遥感领域常用的面向对象图像分割技术引入语谱图分割,并提出了适合于自然界鸟类生物多样性提取的知识规则和斑块统计分析方法。研究实验在杭州植物园的不同区域布点采集音频数据,研究结果表明:鸟类的多样性与地物景观类型和人类活动影响程度密切相关,鸟类叫声的中心频率集中分布在2.5—4.5k Hz之间,最低频率分布在0.67—2.1k Hz之间,最高频率分布在7.6—8.9k Hz之间,人类活动较多的区域,鸟类活动较少且叫声更短促。此外,该方法提取的斑块面积周长比参数,可以定量的反映鸟类叫声的婉转程度。     

Distress calls reflect poxvirus infection in lesser short-toed lark Calandrella rufescens

Several studies have highlighted the association between birdsong and parasite load, but there is no evidence regarding therelationships among pathogens and alarm or distress calls, whichare used in antipredator strategies. We analyzed the associationbetween virus infection and the distress calls of lesser short-toedlark (Calandrella rufescens), addressing the relationships betweencall acoustic properties, presence of poxvirus lesions, andother measurements of bird health (body condition, T-cell–mediatedimmune response, heterophils-to-lymphocytes ratio [H/L ratio],and blood parasites). The study was carried out in Fuerteventura(Canary Islands), where 55% of lesser short-toed larks wereinfected. Pox infection was associated with changes in the spectrotemporalstructure of lark distress calls and affected the condition-dependentnature of these signals. Virus bearers uttered significantlyshorter and lower pitched calls than virus-free birds, givingeven shorter calls when subject to greater physiological stress(higher H/L ratio), whereas virus-free individuals did the opposite.Call harshness was positively correlated with bird immune andbody condition, independently of virus infection, whereas callpulse rate decreased in stressed birds. We hypothesize thathealthy birds might reveal their ability to bear the costs ofantipredator defense by means of long, harsh, and fast modulateddistress calls. Infection not only affects individual stateand morphology by means of gross lesions but also may altera signal used in an antipredator context, thus potentially reducingbird fitness both directly (through disease) and indirectly(through increased predation).     

Performance of autonomous recorders to detect a cryptic and endangered primate species,the black lion-tamarin (Leontopithecus chrysopygus)

Information about species distribution is important for conservation but the monitoring of populations can demand a high sampling effort with traditional methods (e.g., line transects, sound playback) that are poorly efficient for cryptic primates, such as the black lion tamarin (Leontopithecus chrysopygus). Here we investigated the effectiveness of passive acoustic monitoring (PAM) as an alternative method to identify the presence of vocalizing lion tamarins in the wild. We aimed to: (1) determine the maximum distance at which autonomous recorders (Song Meter 3) and Raven Pro acoustic software can respectively detect and identify lion tamarin long calls emitted by two captive subjects (ex situ study); and (2) determine the sampling effort required to confirm the presence of the species in the wild (in situ study). In captive settings, we recorded lion tamarin long calls with one to two autonomous recorders operating at increasing distances from the animals' enclosure (8−202 m). In a 515 ha forest fragment, we deployed 12 recorders in a grid, 300 m apart from each other, within the estimated 100 ha home range of one group, and let them record for 10 consecutive days, totaling 985 h. In the ex situ study, hand-browsing of spectrograms yielded 298 long calls emitted from 8 to 194 m, and Raven's Template Detector identified 54.6% of them, also emitted from 8 to 194 m. In the in situ study, we manually counted 1115 long calls, and the Raven's Template Detector identified 44.75% of them. Furthermore, the presence of lion tamarins was confirmed within 1 day using four randomly sorted recorders, whereas 5 days on average were necessary with only one device. While specific protocols still need to be developed to determine primate population size using this technology, we concluded that PAM is a promising tool when considering long term costs and benefits.     

Wild Female Olive Baboons Adapt their Grunt Vocalizations to Environmental Conditions

Sound propagates differently and visibility varies according to the habitat type. Animals should therefore adapt the acoustic structure and the usage of their vocal signals to the environment. In the present study, we examined the influence of the habitat on the vocal behaviour of wild olive baboons ( Papio hamadryas anubis ) in two populations: one living in Gashaka-Gumti National Park, Nigeria, and the other in Budongo Forest, Uganda. We investigated whether female baboons modified the acoustic structure of their grunts and their rate of grunting when they wandered between closed and open habitat types. As an adaptation to the environmental conditions, baboons might utter calls with a longer duration, a lower fundamental frequency and/or energy concentrated in lower frequencies in a closed habitat like forest than in an open habitat. Baboons should also grunt more frequently in the closed habitat. Analyses showed that in both populations grunts uttered in forest were significantly longer than in open habitat. Additionally, baboons from Uganda showed a significantly higher grunt rate in forest than in open habitat. These results revealed a certain degree of plasticity in vocal production and call usage with regard to the habitat type. However, results in Nigeria suggested that, besides habitat structure, other proximate factors like the context of calling and the proximity between group members could also have an influence on the actual communication patterns.     

虎皮鹦鹉声行为的研究

本文对虎皮鹦鹉(Melopsittacus undelafus)不同声行为叫声的声学特性进行了定量分析,为地震前鸟类声行为的定量观测和鸟声学研究提供了新的资料和认识。 虎皮鹦鹉求偶、离群、受伤及其同伴叫声中变音调声段的声学特性有明显的差异。但是抗议叫声具有鸟类警戒叫声的一般特征,即都有时间图样基本相同的音节组成,每个音节含有若干个调幅脉冲列,其频带和声强都明显加宽和增大。 雏鸟不同的声行为叫声都是简单结构。与成鸟叫声相比较,尤其是抗议叫声呈现明显的发育过程。     

Echolocation calls in Central American emballonurid bats: signal design and call frequency alternation

In southern Central America, 10 species of emballonurid bats occur, which are all aerial insectivores: some hunt flying insects preferably away from vegetation in open space, others hunt in edge space near vegetation and one species forages mainly over water. We present a search call design of each species and link signal structure to foraging habitat. All emballonurid bats use a similar type of echolocation call that consists of a central, narrowband component and one or two short, frequency-modulated sweeps. All calls are multi-harmonic, generally with most energy concentrated in the second harmonic. The design of search calls is closely related to habitat type, in particular to distance of clutter. Emballonurid bats foraging in edge space near vegetation and over water used higher frequencies, shorter call durations and shorter pulse intervals compared with species mostly hunting in open, uncluttered habitats. Peak frequency correlated negatively with body size. Regular frequency alternation between subsequent calls was typical in the search sequences of four out of 10 species. We discuss several hypotheses regarding the possible role of this frequency alternation, including species identification and partitioning of acoustic channels. Furthermore, we propose a model of how frequency alternation could increase the maximum detection distance of obstacles by marking search calls with different frequencies.     

The Use of Automated Bioacoustic Recorders to Replace Human Wildlife Surveys: An Example Using Nightjars

To be able to monitor and protect endangered species, we need accurate information on their numbers and where they live. Survey methods using automated bioacoustic recorders offer significant promise, especially for species whose behaviour or ecology reduces their detectability during traditional surveys, such as the European nightjar. In this study we examined the utility of automated bioacoustic recorders and the associated classification software as a way to survey for wildlife, using the nightjar as an example. We compared traditional human surveys with results obtained from bioacoustic recorders. When we compared these two methods using the recordings made at the same time as the human surveys, we found that recorders were better at detecting nightjars. However, in practice fieldworkers are likely to deploy recorders for extended periods to make best use of them. Our comparison of this practical approach with human surveys revealed that recorders were significantly better at detecting nightjars than human surveyors: recorders detected nightjars during 19 of 22 survey periods, while surveyors detected nightjars on only six of these occasions. In addition, there was no correlation between the amount of vocalisation captured by the acoustic recorders and the abundance of nightjars as recorded by human surveyors. The data obtained from the recorders revealed that nightjars were most active just before dawn and just after dusk, and least active during the middle of the night. As a result, we found that recording at both dusk and dawn or only at dawn would give reasonably high levels of detection while significantly reducing recording time, preserving battery life. Our analyses suggest that automated bioacoustic recorders could increase the detection of other species, particularly those that are known to be difficult to detect using traditional survey methods. The accuracy of detection is especially important when the data are used to inform conservation.     

Low-cost acoustic design of a bat test room

Experiments with captive bats need a flight room that is acoustically neutral, especially when recording and analysing bat calls or the response of bats to certain sound stimuli. Our aim was to identify an isolation material with the best quality–price relationship to acoustically coat such a flight room. For this, we built a flight room divided into two compartments that were to be acoustically isolated from one another. Audible and infrasonic waves are difficult to attenuate with low-cost materials but the attenuation of ultrasounds is rather straightforward. We evaluated the absorbing capacities of different low-cost materials – felt fabric, polystyrene, egg boxes, egg boxes coated with felt fabric, absorbing pyramidal foams, polyurethane foams and cork. The material that showed the best quality–price relationship was the polyurethane foam of open cells (5 cm thickness), which was able to attenuate approximately 20 dB at ultrasonic frequencies.     

绿背山雀繁殖期鸣声声谱分析

2005年4～6月,在四川屏山县老君山自然保护区采用计算机声谱分析技术,对绿背山雀繁殖期护域、警告、报警、警戒和幼雏乞食的鸣声进行了初步研究。结果表明,雄性护域有3种不同音节鸣声,3种音节鸣声的MPF差异均不明显（P〉0.05）,全句持续时间差异显著（X^2=8,42〉x0.05^2,df=2,P〈0．05）;雌性护幼为2个音节的警告呜声;雌雄两性遇到危险时均发出音节不同的警戒声和报警声,其警戒声各音节持续时间差异极显著（P〈0,01）,各音节间隔时间差异不明显（P〉0.05）;幼雏乞食鸣声通常由2～3个音节重复而成。同时还探讨了不同鸣声与相应行为的关系。