Assessing the movement patterns and key habitat features of breeding humpback whales is a prerequ... more Assessing the movement patterns and key habitat features of breeding humpback whales is a prerequisite for the conservation management of this philopatric species. To investigate the interactions between humpback whale movements and environmental conditions off Madagascar, we deployed 25 satellite tags in the northeast and southwest coast of Madagascar. For each recorded position, we collated estimates of environmental variables and computed two behavioural metrics: behavioural state of 'transiting' (consistent/directional) versus 'localized' (variable/non-directional), and active swimming speed (i.e. speed relative to the current). On coastal habitats (i.e. bathymetry < 200 m and in adjacent areas), females showed localized behaviour in deep waters (191 ± 20 m) and at large distances (14 ± 0.6 km) from shore, suggesting that their breeding habitat extends beyond the shallowest waters available close to the coastline. Males' active swimming speed decreased in ...
Although mammalian vocalizations are predominantly harmonically structured, they can exhibit an a... more Although mammalian vocalizations are predominantly harmonically structured, they can exhibit an acoustic complexity with nonlinear vocal sounds, including deterministic chaos and frequency jumps. Such sounds are normative events in mammalian vocalizations, and can be directly traceable to the nonlinear nature of vocal-fold dynamics underlying typical mammalian sound production. In this study, we give qualitative descriptions and quantitative analyses of nonlinearities in the song repertoire of humpback whales from the Ste Marie channel (Madagascar) to provide more insight into the potential communication functions and underlying production mechanisms of these features. A low-dimensional biomechanical modeling of the whale's U-fold (vocal folds homolog) is used to relate specific vocal mechanisms to nonlinear vocal features. Recordings of living humpback whales were searched for occurrences of vocal nonlinearities (instabilities). Temporal distributions of nonlinearities were ass...
The Journal of the Acoustical Society of America, 2015
Automatic transcription of music is a long-studied research field with many operational systems a... more Automatic transcription of music is a long-studied research field with many operational systems available commercially. In this paper, a generic transcription system able to host various prior knowledge parameters has been developed, followed by an in-depth investigation of their impact on music transcription. Explicit links between musical knowledge and algorithmic formalism have been made. Musical knowledge covers classes of timbre, musicology, and playing style of an instrument repertoire. An evaluation sound corpus gathering musical pieces played by human performers from three different instrument repertoires, namely, classical piano, steel-string acoustic guitar, and the marovany zither from Madagascar, has been developed. The different components of musical knowledge have been successively incorporated in a complete transcription system, consisting mainly of a Probabilistic Latent Component Analysis algorithm post-processed with a Hidden Markov Model, and their impact on transcription results have been comparatively evaluated.
... based on the time-varying Schur coefficients for underwater signals analysis Maciej Lopatka1,... more ... based on the time-varying Schur coefficients for underwater signals analysis Maciej Lopatka1,2, IEEE student member, Olivier Adam1, Christophe ... 11, NO.9, September 2004 [4] GMatz, A. Raidl, &quot;Robust detection of nonstationary random signals belonging to p-point uncertainty ...
IEEE/SP 13th Workshop on Statistical Signal Processing, 2005, 2005
This paper proposes a non-stationary time-series segmentation method based on the analysis of the... more This paper proposes a non-stationary time-series segmentation method based on the analysis of the forward prediction error issued from the adaptive Schur orthogonal signal parameterisation. There is no a priori information about the analysed signal thus this method can be easily adapted to a large family of different types of signals for which two different stochastic processes are present. In this paper we set out some of the advantages of the adaptive Schur filter in deducing the presence of different non-stationary transient or long-term events leading to the signal segmentation. For each sample, the adaptive Schur algorithm calculates the optimal second-order solution for the signal prediction resulting in a set of time-varying model parameters (inter alia forward prediction error). We define the likelihood ratio (LR) test based on the Schur forward prediction error that is evaluated at each sample, thus giving excellent time-reaction properties. The LR test allows us to effectively partition the analysed time-series into homogeneous segments by considering its second-order statistics which are tracked adaptively by the Schur filter. The results performed by applying the proposed method to simulated signals are shown to verify its high performance
La population de baleine à bosse de l'ouest de l'océan Indien a été appelée stock C par l... more La population de baleine à bosse de l'ouest de l'océan Indien a été appelée stock C par la Commission Baleinière Internationale (CBI, http://iwc.int/). Pendant la période de reproduction, son aire de distribution se dessine de la côte est de l'Afrique à l'archipel des Mascareignes (Report of the Scientific Committee of the IWC, 1998). Ce stock se divise en 3 sous-régions (Best et al., 1998) : C1 (côte est de l'Afrique du Sud au Mozambique), C2 (canal du Mozambique jusqu'à l'archipel des Comores), C3 (eaux côtières de Madagascar). A ces trois sous-régions, a été proposée une 4 ème qui comprend l'île de la Réunion et l'île Maurice (IWC, 2008a). La dynamique, les échanges et le chevauchement entre ces sous-stocks sont complexes et mal compris (Cerchio et al., 2008; Fleming and Jackson, 2011). La proximité géographique et les échanges observés entre ces différentes régions remettent en cause aujourd'hui la délimitation de ces sous-stocks (Rosenbau...
Using a neural network to identify models and predict signals allows to go beyond the linear doma... more Using a neural network to identify models and predict signals allows to go beyond the linear domain. In this paper, we show the advantage of using neural network for these signal processing applications. Thus, the function charactering the cell (sigmoïd or others) allows the study of non-linear models. Using feedback links specific to a recurrent network, the time is taken
Sperm whales (Physeter macrocephalus) make series of transient echolocation sounds (clicks), when... more Sperm whales (Physeter macrocephalus) make series of transient echolocation sounds (clicks), when foraging. Clicks reflect to the sea surface and the seafloor when propagating towards a receiver. The detection of the once reflected surface and seafloor echoes of a click, and the measurement of the delays of both echoes to the direct path transmitted signal, make possible the localization in depth and in range of the sperm whale which emitted the click, by using a single receiver. Repeating this process click after click then leads to the plotting of the whale depth/range variations while diving. The main difficulty when automating this process is usually to correctly identify direct path signals and echoes. The authors compute the a priori probability density functions of both surface/floor echo delays, improving this click echo identification process. The authors then use an adaptive filter, detecting simultaneously direct path signals with surface echoes. A visual estimation of the range of the whale when starting a dive is used to initiate the detection process. The detection of seafloor echoes then leads to the depth/range estimation of the whale while diving. The detection process works fine (84%, considering a 45-minute dive recording) assuming that a single sperm whale is clicking. The localization process works correctly when seafloor echoes are detected. This method is non-invasive, as it uses neither tags nor active acoustics, and easy to set, as requiring a single hydrophone and a CTD meter, and makes possible an unbiased, inexpensive, automated survey of the diving behavior of single sperm whales in a given area.
ABSTRACT Two new theoretical roles for the mysticete laryngeal sac are proposed here, supported b... more ABSTRACT Two new theoretical roles for the mysticete laryngeal sac are proposed here, supported by observations of a live humpback whale. One role is to regulate the internal air pressure in the respiratory tract independently of ambient pressure, and another one is to control the whale’s buoyancy.
Assessing the movement patterns and key habitat features of breeding humpback whales is a prerequ... more Assessing the movement patterns and key habitat features of breeding humpback whales is a prerequisite for the conservation management of this philopatric species. To investigate the interactions between humpback whale movements and environmental conditions off Madagascar, we deployed 25 satellite tags in the northeast and southwest coast of Madagascar. For each recorded position, we collated estimates of environmental variables and computed two behavioural metrics: behavioural state of 'transiting' (consistent/directional) versus 'localized' (variable/non-directional), and active swimming speed (i.e. speed relative to the current). On coastal habitats (i.e. bathymetry < 200 m and in adjacent areas), females showed localized behaviour in deep waters (191 ± 20 m) and at large distances (14 ± 0.6 km) from shore, suggesting that their breeding habitat extends beyond the shallowest waters available close to the coastline. Males' active swimming speed decreased in ...
Although mammalian vocalizations are predominantly harmonically structured, they can exhibit an a... more Although mammalian vocalizations are predominantly harmonically structured, they can exhibit an acoustic complexity with nonlinear vocal sounds, including deterministic chaos and frequency jumps. Such sounds are normative events in mammalian vocalizations, and can be directly traceable to the nonlinear nature of vocal-fold dynamics underlying typical mammalian sound production. In this study, we give qualitative descriptions and quantitative analyses of nonlinearities in the song repertoire of humpback whales from the Ste Marie channel (Madagascar) to provide more insight into the potential communication functions and underlying production mechanisms of these features. A low-dimensional biomechanical modeling of the whale's U-fold (vocal folds homolog) is used to relate specific vocal mechanisms to nonlinear vocal features. Recordings of living humpback whales were searched for occurrences of vocal nonlinearities (instabilities). Temporal distributions of nonlinearities were ass...
The Journal of the Acoustical Society of America, 2015
Automatic transcription of music is a long-studied research field with many operational systems a... more Automatic transcription of music is a long-studied research field with many operational systems available commercially. In this paper, a generic transcription system able to host various prior knowledge parameters has been developed, followed by an in-depth investigation of their impact on music transcription. Explicit links between musical knowledge and algorithmic formalism have been made. Musical knowledge covers classes of timbre, musicology, and playing style of an instrument repertoire. An evaluation sound corpus gathering musical pieces played by human performers from three different instrument repertoires, namely, classical piano, steel-string acoustic guitar, and the marovany zither from Madagascar, has been developed. The different components of musical knowledge have been successively incorporated in a complete transcription system, consisting mainly of a Probabilistic Latent Component Analysis algorithm post-processed with a Hidden Markov Model, and their impact on transcription results have been comparatively evaluated.
... based on the time-varying Schur coefficients for underwater signals analysis Maciej Lopatka1,... more ... based on the time-varying Schur coefficients for underwater signals analysis Maciej Lopatka1,2, IEEE student member, Olivier Adam1, Christophe ... 11, NO.9, September 2004 [4] GMatz, A. Raidl, &quot;Robust detection of nonstationary random signals belonging to p-point uncertainty ...
IEEE/SP 13th Workshop on Statistical Signal Processing, 2005, 2005
This paper proposes a non-stationary time-series segmentation method based on the analysis of the... more This paper proposes a non-stationary time-series segmentation method based on the analysis of the forward prediction error issued from the adaptive Schur orthogonal signal parameterisation. There is no a priori information about the analysed signal thus this method can be easily adapted to a large family of different types of signals for which two different stochastic processes are present. In this paper we set out some of the advantages of the adaptive Schur filter in deducing the presence of different non-stationary transient or long-term events leading to the signal segmentation. For each sample, the adaptive Schur algorithm calculates the optimal second-order solution for the signal prediction resulting in a set of time-varying model parameters (inter alia forward prediction error). We define the likelihood ratio (LR) test based on the Schur forward prediction error that is evaluated at each sample, thus giving excellent time-reaction properties. The LR test allows us to effectively partition the analysed time-series into homogeneous segments by considering its second-order statistics which are tracked adaptively by the Schur filter. The results performed by applying the proposed method to simulated signals are shown to verify its high performance
La population de baleine à bosse de l'ouest de l'océan Indien a été appelée stock C par l... more La population de baleine à bosse de l'ouest de l'océan Indien a été appelée stock C par la Commission Baleinière Internationale (CBI, http://iwc.int/). Pendant la période de reproduction, son aire de distribution se dessine de la côte est de l'Afrique à l'archipel des Mascareignes (Report of the Scientific Committee of the IWC, 1998). Ce stock se divise en 3 sous-régions (Best et al., 1998) : C1 (côte est de l'Afrique du Sud au Mozambique), C2 (canal du Mozambique jusqu'à l'archipel des Comores), C3 (eaux côtières de Madagascar). A ces trois sous-régions, a été proposée une 4 ème qui comprend l'île de la Réunion et l'île Maurice (IWC, 2008a). La dynamique, les échanges et le chevauchement entre ces sous-stocks sont complexes et mal compris (Cerchio et al., 2008; Fleming and Jackson, 2011). La proximité géographique et les échanges observés entre ces différentes régions remettent en cause aujourd'hui la délimitation de ces sous-stocks (Rosenbau...
Using a neural network to identify models and predict signals allows to go beyond the linear doma... more Using a neural network to identify models and predict signals allows to go beyond the linear domain. In this paper, we show the advantage of using neural network for these signal processing applications. Thus, the function charactering the cell (sigmoïd or others) allows the study of non-linear models. Using feedback links specific to a recurrent network, the time is taken
Sperm whales (Physeter macrocephalus) make series of transient echolocation sounds (clicks), when... more Sperm whales (Physeter macrocephalus) make series of transient echolocation sounds (clicks), when foraging. Clicks reflect to the sea surface and the seafloor when propagating towards a receiver. The detection of the once reflected surface and seafloor echoes of a click, and the measurement of the delays of both echoes to the direct path transmitted signal, make possible the localization in depth and in range of the sperm whale which emitted the click, by using a single receiver. Repeating this process click after click then leads to the plotting of the whale depth/range variations while diving. The main difficulty when automating this process is usually to correctly identify direct path signals and echoes. The authors compute the a priori probability density functions of both surface/floor echo delays, improving this click echo identification process. The authors then use an adaptive filter, detecting simultaneously direct path signals with surface echoes. A visual estimation of the range of the whale when starting a dive is used to initiate the detection process. The detection of seafloor echoes then leads to the depth/range estimation of the whale while diving. The detection process works fine (84%, considering a 45-minute dive recording) assuming that a single sperm whale is clicking. The localization process works correctly when seafloor echoes are detected. This method is non-invasive, as it uses neither tags nor active acoustics, and easy to set, as requiring a single hydrophone and a CTD meter, and makes possible an unbiased, inexpensive, automated survey of the diving behavior of single sperm whales in a given area.
ABSTRACT Two new theoretical roles for the mysticete laryngeal sac are proposed here, supported b... more ABSTRACT Two new theoretical roles for the mysticete laryngeal sac are proposed here, supported by observations of a live humpback whale. One role is to regulate the internal air pressure in the respiratory tract independently of ambient pressure, and another one is to control the whale’s buoyancy.
Uploads