Why Write a Bioacoustics Methods eBook?
The intent of this volume is twofold: (a) to promote multidisciplinary and comparative analyses of sound production, propagation, and perception in biological organisms and (b) provide a source of relevant methods for the novice bioacoustician. The first goal of this initiative was approached by crossing disciplinary boundaries. The exchange between scientists in biology, psychology, neuroscience, engineering, anthropology, speech, voice, hearing, and related sciences appears critical for bioacousticians. Over the course of our careers we have collaborated with personnel trained in a wide variety of disciplines including radio broadcast engineers, speech and hearing scientists, psychologists, physiologists, anatomists, epidemiologists, ethologists, anthropologists, physicists, electrical engineers, veterinarians, zoologists, zoo keepers, vocal musicians, and sound archivists. We have collaborated with faculty members, postdoctoral fellows, as well as graduate and undergraduate students. All of these individuals brought a fresh and unique perspective to the study of bioacoustics; they all had a domain of expertise in some area of bioacoustics, and they all exhibited areas of limited familiarity to topics (and methodologies) of interest to other members of the bioacoustics community. Furthermore, the wealth of research methods available, in many instances dictates the collaborative approach.
Vocal communication is undoubtedly one of the most fascinating and complex behaviors that animals perform. It requires the analysis at all four levels (mechanism, adaptive significance, development and evolution) identified by Tinbergen (Bateson, Laland, 2013). Historically, bioacoustics research started as a small sub-discipline in ethology (e.g., Tembrock, 1959; Busnel, 1963; Sebeok, 1977; Nikolskii, 1984). Over the last five decades not only has an enormous amount of knowledge been accumulated (e.g., Kroodsma, Miller, 1996; Gerhardt, Huber, 2002; Bradbury, Vehrencamp, 2011; Hopp et al., 2012; Hedwig, 2014; Wiley, 2015), but new methods have been employed or developed which require more know-how than just operating a handheld microphone and tape recorder and subsequent acoustic analysis in the temporal and spectral domain. In light of the second goal i.e. providing a source for relevant methods, this volume can represent only a beginning. The intent of this book was also to initiate the creation of a repository for the research methodologies used by bioacousticians. The book may serve as a resource intended to help members of the bioacoustics community to communicate more skillfully with one another, to serve as a reference for training our students, to promote greater collaboration, to recruit new investigators, and to help investigators adopt new techniques to strengthen their research programs. In this respect several years ago I (CB) was contacted by Asif Ghazanfar of the Department of Psychology at Princeton University inquiring if there was a source that would essentially “coach” a graduate student or faculty member how to set up an excised larynx bench for their laboratory. We observed that there was no bioacoustical methods publication that was intended to instruct members of the research community how to learn a new technique or initiate a research project in an area in which they had no prior exposure or training, but there should be! Asif’s inquiry is one of the key exchanges we had with our colleagues over the past several years that served as inspiration for developing this methods eBook.
Comparative bioacoustics is extraordinarily broad in scope. It includes the study of sound propagation, dispersion, attenuation, absorption, reverberation, and signal degradation; as well as sound detection, recognition, and classification in both marine and terrestrial organisms (including humans). This research is informed by an understanding of the mechanisms underlying sound generation and aural reception, as well as the anatomy and physiology of the organs dedicated to these functions, and their ontogeny and development. Furthermore, it includes studies of the ontogeny of vocal behavior and the relationship between the environment and the acoustic behaviors of all conspecific organisms resident in the local biome. Comparative bioacoustic researchers have developed signal processing algorithms for taxonomic classification of the acoustically conspicuous biota in the habitat, and for the calculation of biodiversity. It includes studies of the effects of industrial noise on the integrity of urban, suburban and rural terrestrial environments, and the impact of man-made noise on marine organisms. It includes the application of acoustic instrumentation for pest-control, the abatement of collisions with wildlife, and wildlife monitoring. The frequency range of the acoustic signals under consideration is equally broad. It includes studies of echolocation, and the utilization of infrasonic, sonic and ultrasonic signals.
Last but not least, the ebook format provided a unique opportunity for the development of a multimedia methods book. As bioacousticians we publish sound spectrograms, and descriptions of audio signals. We conduct measurements of the acoustic profile of a variety of habitats, and we measure and describe the mechanisms of animal sound production, but the acoustic signals that are the focus of our field are inaccessible to our readership. A key feature of this volume is our attempt to bring the field of bioacoustics fully into the 21st Century by promoting the expectation of the integration of audio and video clips into a scholarly text. In this volume readers have the ability to “hear” associated audio files by “clicking” on the speaker icon linked to a sound spectrogram or to the text itself. Where appropriate, short video clips have also been included. One goal of this volume has been to capitalize on recent developments in digital publishing in order to improve dissemination of information concerning the central methods and findings in the field of bioacoustics. One long-term goal of this effort is to encourage the creation of scholarly publications in which scientists are encouraged, and potentially expected, to archive rare audio recordings in association with their notes, measurements and observations. Audio archiving should be granted full recognition as an important component of scholarship in bioacoustics, and archived audio files give the scientific community an opportunity to study how habitats change over time in terms of their acoustic properties as the world experiences a change in climate, shifts in the density of human habitation, invasion of exotic species and so forth. For the field of bioacoustics, the benefit of such archiving is obvious and this volume strives to elevate the recognition of scholarship associated with the archiving enterprise.
An Overview of this Volume
This methods book would be encyclopedic if it exhaustively addressed all the domains encompassed by comparative bioacoustics. This eBook is merely a starting point for building a methodology repository. We hope that it will lead to subsequent editions that up-date and expand the methodologies presented here. Because this book is electronic where appropriate we include both video and audio files to illustrate the author’s key points. This book is partitioned into four sections. Part 1 is composed of two chapters laying the ground work for an understanding of acoustics. Chapter 1 by Larsen and Wahlberg is about the physics of sound in air and how it is produced. Bioacousticians handle sound in four major contexts: sound recording, sound analysis, sound synthesis, and sound playbacks. The text is designed to achieve a “sweet spot”, so that it is intended to be appropriate for novices (such as advanced undergraduate students), but to retain sufficient rigor so that the treatment is appropriate for students seeking preparation for a scholarly career in some aspect of comparative bioacoustics. The topics covered include the nature of sound waves in air, sound pressure, intensity and power. The reader is introduced to the concept of sound fields and sound propagation. The relationship between the period of the wave and wavelength, the concept of acoustic impedance, the acoustic near field and the acoustic far field are discussed. The reader is introduced to sound calibration, and the differences between a diffuse sound field, a semi-reverberant field, and a closed sound field. Frequently novices are unaware of phenomena like diffraction, the concept of an acoustic monopole, acoustic dipole, and the problem of destructive and constructive interference, and their treatment in this chapter is designed to help fledgling bioacousticians think about biological sound sources, and sound measurement with greater sophistication. The last section in this chapter considers the idea that it may be desirable for animals to beam or direct their signals towards an intended recipient, the design of a biological acoustic horn, and the strategies that investigators may employ to strive to measure the directional properties of animal acoustic displays is discussed.
Wahlberg and Larsen expand their treatment on the physics of sound and sound sources in chapter 2. This chapter is devoted to an in depth treatment of sound propagation both in air and in water. In nature the sound wave incident at the receiver is reduced in amplitude relative to the emitted signal, and more importantly the path from sender to recipient is rarely straight, and the received signal is typically distorted, degraded or blurred due to frequency dependent absorption of the sound wave by the medium within which it is propagated, and also by scattering, reflection and absorption of the signal with large and small objects encountered in the environment during its propagation. The reader is introduced to the ideas of geometric attenuation, excess attenuation, absorption by the medium and sound propagation from one medium to another. In the natural world the media is rarely homogeneous. Both meteorological and underwater conditions change the velocity of sound propagation, and the media becomes stratified causing refraction or curvature in the acoustic ray. To complicate matters further the presence of water currents and wind in the atmosphere cause turbulence in the media. The media moves in loops or eddies producing distortions of the wavefront. The efficacy of both echolocation systems and acoustic communication systems is dependent upon specializations in signal emission strategies, or perceptual processing strategies devoted to the task of assessing the likelihood that differences between two successive waveforms incident at the receiver are due to instabilities in the propagation of the signal in the habitat, or if these differences are due to changes in the waveform reflected from the target in the case of echo location systems, or are due to intended differences in the “design” of the emitted signal in the case of acoustic communication systems. These issues are of fundamental importance for many questions addressed in comparative bioacoustics.
Part 2 is composed of three chapters which focus on vocal production in terrestrial vertebrates. Chapter 3 by Brown and Riede reviews the biomechanics of sound production by the larynx. The myoelastic-aerodynamic theory of voice production accounts for most of the vocalizations produced by frogs, reptiles, and mammals. Accordingly, sound is generated by a repeating cycle in which the glottis opens and closes disrupting or modulating the transglottic air flow. The larynx with a small range in variation in physical size is remarkable in its capacity to allow for a very broad range in the fundamental frequency of oscillation both within a species, and across taxa. This phenomenon is dependent upon the viscoelastic properties of the superficial layers of the vocal folds, the layered composition of the vocal folds, and the capacity to selectively adjust the stiffness of the tissue layers within the folds. The oscillation of the vocal folds behaves almost like a string. Stretching the vocal folds causes the tissue to vibrate at higher rates as the tissue becomes stiffened much like a string. Comparative mammalian data show that there are significant species differences in the composition of the vocal folds that should impact on their viscoelastic properties. Species differ with respect to the number of layers of the lamina propria within the vocal folds, and the distribution and density of fibrous proteins, hyaluronan and fat cells. Differences in the composition of the vocal folds will alter their biomechanical properties, and the biomechanical properties of the vocal folds can be measured by two complimentary experimental procedures. In the first procedure the vocal folds are dissected from the larynx, and tissue stress is calculated using force and length measurements. This procedure requires prior macroscopic and microscopic investigation of the morphology of the vocal folds for the species in question so that the tissue is dissected appropriately. The excised larynx procedure is the second experimental manipulation commonly used to explore species differences in the biomechanics of the larynx. In this preparation the laryngeal complex is mounted on a hollow tube, and the passage of air through a “pseudo-trachea” initiates oscillations of the vocal folds much like during natural vocal behavior.
The syrinx, the avian vocal organ is a unique sound source among tetrapods. Chapter 4 by Goller reviews sound production and modification in birds, and the methodology employed to study the vocal behavior of birds. Birds are equipped with one or two sound sources within the syrinx. In songbirds, the left and right syrinx are controlled independently permitting the possibility of simultaneous or asynchronous phonation. There is substantial morphological variation in the shape of the cartilages, muscles and membranes and labia in the syrinx between different families of birds, and we begin slowly to make sense of how the morphological variability facilitates a species’ typical vocal repertoire. Investigating sound production in birds is a challenge because structures are very small. Goller describes a number of techniques which have been developed. For example, airflow has been measured in spontaneously singing birds with small microbead thermistors implanted in the airway. Fine wire EMG electrodes permit recording the activity of syringeal muscles, and the nerves innervating these muscles can be severed to observe their role in the control of singing. After sound is generated by the syrinx, it travels through a vocal tract before radiated from the beak. The suprasyringeal vocal tract filters or shapes the sound generated in the syrinx. Some aspects of the vocal tract are highly dynamic, including the beak and hyoid movements which have been studied by x-ray filming.
The idea that the geometry of the airway shapes the sound produced by the syrinx or larynx is known as the source-filter theory. This theory is addressed by Hunter and Ludwigsen in chapter 5. The authors introduce essential concepts of acoustic filters and resonances beginning with two archetypes of acoustic filters: the Helmholtz resonator and the pipe resonator. Knowing the dimensions of the respective air filled cavity allows acousticians to make fairly precise predictions about its resonance characteristics. These calculations are explained in simple mathematical terms for the Helmholtz and the pipe resonator. The text also presents an approach which allows investigators to empirically determine the resonance or filter properties of an organism’s airway. The organism’s filter is acoustically excited either by one of three signals: a swept sine waves, white noise or by a pulse. This allows the demonstration of resonance frequencies in the frequency spectrum of the radiated sound. Finally, the authors apply these concepts to human and animal vocalizations and demonstrate how the properties of the source and filter can be extracted using spectrograms and power spectra. The sounds we deal with are most often those radiated from the lips of a speaking human or animal subject or from the beak of a bird. The sound therefore contains features that pertain to source characteristics and others that can be related to filter properties. To fully understand communication signals, bioacousticians must consider the properties of the source, and the properties of the filter, both parameters reveal relevant information about the sender. However, as noted in chapter 5, it is also important to understand the limitations and trade-offs of a source-filter approach to the acoustic analysis of vocal production.
Part 3 is focused on sound analysis methodologies. This section is comprised of five chapters. The topics discussed include sound classification using animal preference methods and computational approaches. The latter encompass sound conditioning, nonlinear dynamics, and sound classification using artificial neural networks and hidden Markov model methodologies.
Clearly bioacousticians cannot depend solely on the output of automated sound classification systems, and additional methods are needed to validate the output of software classification systems, to determine how well they line up with the perceptual judgments and preferences of living organisms. A variety of behavioral research methods focused on acoustic preference testing are described by Riebel in chapter 6. Vocal displays are often used to advertise for a mate, and many species are attracted to mating signals, and the opportunity to hear these calls can be used as a reward or reinforcement in operant conditioning tasks. Acoustic preferences can be measured through a wide range of assays including phonotaxis tests, loudspeaker approach assays using a maze, nest cavity choice tests, copulation solicitation assays, nest building assays, and vocal responsiveness assays. In addition to passive exposure tests, the sound playback can be triggered in operant preference tests using key pecking and perch hopping technologies. Behavioral preference tests can be correlated with physiological measures including heart rate, hormone profiles, maternal resource allocation and fMRI imaging studies. Riebel argues that these methods are not all equally suitable for every research species or research question, and researchers should seek to use a combination of methods to establish the internal and external validity of their data sets.
Chapter 7 by Stoddard and Owren reviews best practices for sound conditioning and preprocessing. The appropriate use of filters is essential before a recorded sound can be analyzed. Filters, either hardware or digital, include band-pass, low-pass, or high-pass versions. They all apply frequency-dependent attenuation to the recording to emphasize a portion of the audio spectrum, remove or minimize noise, and define the upper and lower frequency boundaries of the recording. Filters can be used to search for signals embedded in background noise and improve the signal-to-noise ratio, remove 60 cycle hum or low-frequency wind noise from a recording, and simulate environmental excess attenuation on broadcast signals in nature. The misuse of filters or the failure to use antialising filters during analog to digital conversion can produce strikingly unintended effects, and chapter 7 discusses these issues which are part and parcel to everyday work in a bioacoustics laboratory.
Animal vocalizations are produced by vocal fold oscillations that range from nearly periodic vibrations to noisy signals generated by chaotic aperiodic oscillations. The field of nonlinear dynamics has helped to make sense of what we perceive as voice breaks, rough and creaky voice, or as screams. Video recordings of the patterns of vibration seen in both human and animal vocal folds show several distinctly different patterns of oscillation. In chapter 8, Tokuda introduces the concept of nonlinear attractors and bifurcation in order to explain acoustic phenomena such as subharmonics, biphonation, deterministic chaos, and frequency jumps. These phenomena all occur in normal phonation as well as in disorder voicing. The data reviewed in this chapter shows how the quantification of nonlinear properties of animal vocalizations, and the investigations based on temporal analytic approaches can be extremely fruitful.
The next two chapters introduce methods devoted to software classification of speech and animal vocalizations. Bioacousticians have traditionally sorted animal calls according to subjective perceptual properties, spectrographic features, or statistically analyses of cardinal features such as duration or fundamental frequency. Johnson and Clemins (Chapter 9) provide an overview of a variety of approaches bioacousticians have employed to analyze and classify animal vocalizations. They identify some of the advantages and limitations of statistical models, template matching models, spectrogram cross-correlation methods, and several machine learning pattern recognition approaches suitable for sound classification. One obvious approach is to adopt the signal classification system employed by speech recognition technology to the task of classifying animal vocalizations. Speech recognition technology, available on most smart devices, takes an acoustic input and assigns it to a phoneme and subsequently to a letter in the alphabet. This technology, founded on the Hidden Markov Model (HMM) signal classification system, is described in detail. The HMM is a state machine, it takes acoustic input sequences, and estimates the most likely corresponding state sequences that produced the input. Though developed for the mainstream application of speech recognition, this approach is flexible and it can be adopted for the automated detection of animal vocalizations, species recognition, call-type classification, and individual identification. Throughout this chapter the authors guide the reader on how the HMM system can be applied to detect, sort, cluster, and classify animal sounds. This tool holds promise for monitoring, conducting a vocalization survey, and census on free ranging populations of animals in remote habitats.
Chapter 10 by Mercado and Sturdy describes a second computational method for sorting sound patterns based on artificial neural networks. Computation neural networks are inspired by the kinds of processes that occur within the brain. The output from one unit in a network serves as an input to another, and the outputs can be weighted to simulate variations in the strengths of synapses in brains. Neural networks are programs for recognizing patterns. One major advantage of this approach is that the network permits nonlinear classification of an acoustic dimension that superficially appears to be continuous, and the expression of nonlinear classification resembles categorical processing for some perceptual phenomena. Neural networks hold promise for identifying natural auditory categories that would not be detected by traditional statistical and spectrographic classification analyses. However, the output of the network depends on the initial architecture of the network and the learning algorithm used to “train” the network. Researchers must decide how to select the initial parameters for the network without clear guidelines on how to select the most appropriate architecture for the task at hand. By comparing the classification of sounds by animals with the classification of sounds by neural networks it is possible to determine if the network and animals are processing similar or dissimilar acoustic cues.
Bioacousticians frequently collect an unusual and interesting inventory of animal sounds over the course of their careers, and while their research methods, results and conclusions are preserved in their scholarly publications, the animal signals which are the basis of their inquiries frequently are not. Thus, it is important for members of the comparative bioacoustic community to intermittently take stock of their inventory of recorded sounds and to submit them to a sound archive. Part 4, composed of one chapter, is focused on sound recording and archiving. Webster and Budney in chapter 11 provide guidelines for collecting and identifying sound media specimens, and good practice for recording notes and observations, and the sounds themselves pertinent to their submission to a sound archiving facility. Archived sounds of focal species, and samples of biotic sounds in endangered or threatened habitats provide the bedrock for the next generation of bioacousticians to monitor how well the soundscape has been preserved or changed in relationship to naturally occurring events such as earthquakes, or other events such as climate change, invasive species, and human activities.
Bateson, P., Laland, K.N. (2013). Tinbergen’s four question: an appreciation and an update. TREE 28,
Busnel, R.G. (1963). Acoustic Behavior of Animals.. New York: Elsevier.
Gerhardt, H.C., Huber, F. (2002). Acoustic Communication in Insects and Anurans. Chicago, USA: The
University of Chicago Press.
Bradbury, J.W., Vehrencamp, S.L. (2011). Principles of Animal Communication. (2nd edition.).
Sunderland MA: Sinauer Associates.
Sebeok, T.A. (1977). How Animals Communicate. Bloomington: Indiana University Press.
Kroodsma, D.E., Miller, E.H. (1982). Acoustic Communication in Birds. Vol. 1 and 2. Production,
perception, and design features of sounds. New York: Academic Press.
Hedwig, B. (2014). Insect Hearing and Acoustic Communication. Berlin: Springer.
Hopp, S.L., Owren, M.J., Evens, C.S. (2012). Animal Acoustic Communication. Berlin: Springer.
Nikol'skii, A.A. (1984). Sound signals of Mammals in Evolutionary Process. (p. 199 pp). Moscow,
Nauka: (In Russian)
Tembrock, G. (1959). Tierstimmen. Eine Einführung in die Bioakustik. Wittenberg, Ziemsen Verlag
Wiley, R.H. (2015). Noise Matters: the Evolution of Communication. Cambridge MA: Harvard