342 research outputs found
Datasets of odontocete sounds annotated for developing automatic detection methods
No full textThis report documents the progress to date of this project to (1) collect recordings of beaked whales and other odontocete species, (2) annotate the collected sound files to make them useful to researchers working on automatic call detection and classification, (3) make them publicly available in an archive on the Internet (www.mobysound.org), and (4) use them in developing automatic detection algorithms and software. Among other things, one important development from this project has been the invention of a method known as Energy Ratio Mapping Algorithm (ERMA) for detecting clicks of beaked whales and other odontocetes. A journal paper and numerous presentations about (and using recordings from) this project have been published and presented. Copies of the abstracts of these publications and presentations are included in this report.N00244-07-1-0005
Kyrgyzstan’s Manas epos millennium celebrations: post-colonial resurgence of Turkic culture and the marketing of cultural tourism
No full textThe paper addresses the symbolic nature of the Manas epos and its influence on both the unification of Kyrgyzstan and the enhancement of the country's national and Turkic identity. The case of the Manas epos millennium celebrations event is then used to illustrate the relationship between the uses of the Manas 'legend' in the construction of a national identity and in the positioning of the cultural tourism product. The paper subsequently assess the potential usefulness of the Manas epos in the creation of a destination image for Kyrgyzstan and in the positioning of Kyrgyzstan in the global tourism marketplace
Recommended from our members
Automatic Classification of Ultrasonic Harbor Porpoise Clicks in a Varying Noise Environment
No full textThis study compares three approaches in the design of an autonomous machine listening agent that predicts harbor porpoise ultrasonic echolocation clicks in diverse noise environments. Considering the temporal variations of noisy coastal ocean soundscapes which the harbor porpoises inhabit, we propose a leave-one-day-out (LODO) cross-validation strategy in the training of a random forest classifier that successfully addressed the covariate shift present in our time-series data. To evaluate the efficacy of our approach to capture signals in this noisy environment, we compare three preprocessing approaches and two deep learning architectures on our harbor porpoise click data. We find that feature extraction strategies of mel frequency cepstral coefficients (MFCC) and short time Fourier transform (STFT) outperformed our novel approach, the heterodyned-Teager-Kaiser Energy Operator (TKEO), which shifts down the ultrasonic signal to a lower frequency in the time domain. Building on these results, we seek to improve the robustness of our porpoise click classifier for a real-world environment by implementing a second-stage stacked random forest ensemble on combinations of subsets of 42 deep learning base models that were trained from the folds of our LODO cross-validation and the three preprocessing approaches that were explored in this study. Our results demonstrate that experiments using the LODO cross-validation strategy reported a difference between the average fold accuracy and a held-out test accuracy of 6%, while training without cross-validation and the equal k-fold cross-validation reported a 28.7% and 30.4% difference, respectively. From the three preprocessing approaches we implement, the models trained on MFCC produced the highest accuracy of 95.6% on the held-out test set while those trained on STFT and heterodyned-TKEO produced accuracies on the same held-out test set of 88.7% and 85.0%, respectively. Results from our stacked random forest show the greatest improvement in accuracy of 5.6% in the heterodyned-TKEO models while the STFT and MFCC models reported 4.5% and 1.9% improvements in accuracy, respectively. Highly varying noise environments are common across coastal areas inhabited by harbor porpoises. This study, with our proposed ensemble of different feature and model architectures, emphasizes the need to overcome such shifts in noise to design a robust porpoise click classifier that is ready for real-time deployment and able to generalize to all real-world conditions
Cheap DECAF: Density Estimation for Cetaceans from Acoustic Fixed Sensors Using Separate, Non-Linked Devices
No full text
Datasets of Odontocete Sounds Annotated for Developing Automatic Detection Methods, FY09-10
No full textDetection, classification, and localization (DCL) research on marine mammal vocalizations has been in development for decades, and methods for marine mammal population density estimation using acoustic data have been in development since at least 2007. These efforts have been supported by MobySound, an archive of cetacean sounds used for studying call detection and localization that are annotated to facilitate research in DCL. This project was aimed to begin development of high-performing automatic detection methods for the sounds of beaked whales and other odontocetes. Specifically, this report [1] details the newly collected odontocete recordings that have been added to the MobySound archive; [2] documents continuing development of methods for detection and classification, including improvements to the Energy Ratio Mapping Algorithm (ERMA) method for use on gliders and its extension to new species and populations; [3] reports on development of a new method for estimating the population density of baleen whales using the summed energy in a frequency band in which they vocalize; and [4] also reports on the successful production of datasets focused on odontocete whistles and clicks and baleen whale calls for the Fifth Workshop on Detection, Classification, Localization, and Density Estimation of Marine Mammals using Passive Acoustics.N00244-09-1-007
- …
