Marine Ecoacoustics and Informatics Lab

Studying the dynamics of marine ecosystems through informatics

Welcome to the MEIL based in Biodiversity Research Center, Academia Sinica. Led by Dr. Tzu-Hao Lin, the MEIL investigates the applications of ecological informatics in biodiversity monitoring and conservation management.

Ocean sound is one of our primary subjects. We integrate passive acoustics and machine learning to track the spatial-temporal changes of biotic and abiotic sounds. By characterizing soundscapes, we seek to understand the mutual impacts among marine ecosystems, climate change, and anthropogenic activities.

We also interest in other non-acoustic techniques that can visualize the dynamics of marine ecosystems. Our goal is to generate open data and reusable models that can facilitate the retrieval of ecological information and increase the efficiency of decision making.


Open source toolbox for soundscape information retrieval

soundscape_IR is an open-source Python toolbox dedicated to soundscape information retrieval. This toolbox provides algorithms for supervised and unsupervised source separation (SS). It also enables the use of a snapshot recording for model training and subsequently applying adaptive and semi-supervised SS when target species produce sounds with varying features and when unseen sound sources are encountered.

Now with V1.1 release, you can use a few lines of code to run source separation detector and feature extraction on a large dataset of soundscape recordings.

Check V1.1 full documentation for details!

Building a citizen science platform for listening marine ecosystem health

Marine soundscapes have been considered as a proxy of marine biodiversity. Is it possible to develop a citizen science project based on marine soundscapes? We compared the quality of audio data collected by an underwater sound recorder commonly applied in academic research and a commercially available underwater video recorder. Despite the distortion of spectral features, an underwater video recorder can still capture sounds of soniferous fish and snapping shrimps. Therefore, we propose:

  1. Deploy underwater video recorders on coral reefs or seagrass beds. Away from the recorders for a short period to reduce noise due to diving regulators and bubbles.

  2. Share your video files via cloud storage and upload the metadata to an open repository.

  3. We will establish an open science computation platform to identify sounds of crustaceans, soniferous fish, and anthropogenic noise.

  4. The analysis result will be uploaded to an open repository so that everyone can explore the acoustic diversity among locations!

This is our preliminary idea. Please let us know if you have any suggestions!

Feature Outcomes

Yi-Jen Sun, Shih-Ching Yen, Tzu-Hao Lin (2022) soundscape_IR: A source separation toolbox for exploring acoustic diversity in soundscapes. Methods in Ecology and Evolution, early view.

soundscape_IR is an open-source Python toolbox dedicated to soundscape information retrieval. This toolbox provides algorithms for supervised and unsupervised source separation (SS). It also enables the use of a snapshot recording for model training and subsequently applying adaptive and semi-supervised SS when target species produce sounds with varying features and when unseen sound sources are encountered.

Using two case studies featuring the diversity of sika deer vocalizations and estuary biotic and abiotic sounds, our results demonstrate that soundscape_IR offers a promising method for streamlining the assessment of acoustic diversity in diverse environments. We expect this toolbox can open new directions for ecoacoustic studies in the future.

Miles James Parsons, Tzu-Hao Lin, T. Aran Mooney, Christine Erbe, Francis Juanes, Marc Lammers, Songhai Li, Simon Linke, Audrey Looby, Sophie L Nedelec, Ilse Catharina Van Opzeeland, Craig Aaron Radford, Aaron N. Rice, Laela Sayigh, Jenni Stanley, Edward Urban, Lucia Di Iorio (2022) Sounding the call for a global library of biological underwater sounds. Frontiers in Ecology and Evolution, 10: 810156.

At a time when worldwide biodiversity is in significant decline and underwater soundscapes are being altered as a result of anthropogenic impacts, there is a need to document, quantify, and understand biotic sound sources–potentially before they disappear. A significant step toward these goals is the development of a web-based, open-access platform that provides: (1) a reference library of known and unknown biological sound sources (by integrating and expanding existing libraries around the world); (2) a data repository portal for annotated and unannotated audio recordings of single sources and of soundscapes; (3) a training platform for artificial intelligence algorithms for signal detection and classification; and (4) a citizen science-based application for public users. We discuss the benefits such a program can provide and the challenges that need to be overcome to bring together bio- and ecoacousticians, bioinformaticians, propagation experts, web engineers, and signal processing specialists (e.g., artificial intelligence).

Chong Chen, Tzu-Hao Lin, Hiromi Kayama Watanabe, Tomonari Akamatsu, Shinsuke Kawagucci (2021) Baseline soundscapes of deep-sea habitats reveal heterogeneity among ecosystems and sensitivity to anthropogenic impacts. Limnology and Oceanography, 66: 3714-3727.

We recorded and analyzed the total environmental sound (‘soundscape’) of four deep-sea sites around Japan, including a hydrothermal vent and an abyssal plain site eyed for the extraction of deep-sea mineral resources. Different habitat types exhibited distinct soundscape characteristics, for example the vent site was dominated by sounds from geological activities and the abyssal plain was much quieter than all other sites. Our results indicate offshore mining-targeted areas are more susceptible to impacts from anthropogenic noise, and that the monitoring of soundscapes are important for environmental assessments in the deep ocean.

Tzu-Hao Lin, Tomonari Akamatsu, Yu Tsao (2021) Sensing ecosystem dynamics via audio source separation: A case study of marine soundscapes off northeastern Taiwan. PLoS Computational Biology, 17: e1008698.

This work use a computational approach that features the automatic separation of sound sources from a mixture to visualize the diurnal, lunar, and seasonal patterns of marine mammals, soniferous fishes, and shipping activities from 2.5 years of underwater sound recordings. This technique not just helps us evaluate the phenological patterns of each sound source and their potential interactions, it also facilitates the assessment of acoustic diversity. As a result, we identified five types of marine mammal vocalizations (including echolocation clicks and whistles), two types of fish choruses that were not reported in this region before. By integrating this computational technique in global acoustic observatories, we can investigate the changes in marine biodiversity at various temporal and spatial scales. In addition, we can also use the information specific to anthropogenic noise to monitor human activities and evaluate their impacts on marine biodiversity. We believe that marine soundscapes will be an indispensable component of marine conservation in the future.

Tzu-Hao Lin, Tomonari Akamatsu, Frederic Sinniger, Saki Harii (2021) Exploring coral reef biodiversity via underwater soundscapes. Biological Conservation, 253: 108901.

This work demonstrated the application of using soundscapes in studying the temporal and spatial changes of coral reef soundscapes. The results revealed amazing dynamics of biological sounds in upper-mesophotic coral reefs. High diversity of biological sounds may be an indicator of coral reef health, long-term monitoring of acoustic diversity will help us know how coral reefs respond to natural and anthropogenic stressors. Our results also discovered that mesophotic corals are frequently exposed to anthropogenic noise. Considering that soundscapes are an important habitat cue for larvae of corals and reef-associated fish, noise masking may interfere the phonotaxis and reduce the resilience of coral reefs. We urge to integrate soundscape monitoring in the global research network of coral reefs by characterizing the underwater sounds in healthy and degraded reef habitats, before and after bleaching events and other marine extreme events, in habitats with varying levels of management and regulations. This action will contribute to a data-informed platform for guiding conservation management and strategic investment on regional and global scales.