海洋聲景訊息擷取之技術發展與應用(Applying soundscape information retrieval in underwater remote sensing)
Have you ever imagined what the ocean sounds like? The ocean is full of sounds that are generated from geophysical events, marine
animals, and human activities.  Listening to underwater sounds allows us to remotely acquire ecological data of marine biodiversity
and investigate changes of biodiversity in response to anthropogenic development. In this presentation,
I will introduce how we apply the soundscapes as a remote sensing platform to explore the spatio-temporal patterns of marine
biodiversity and anthropogenic activities.
On the basis of international collaboration, we initiated the Ocean Biodiversity Listening Project that aims to characterize soundscape
features from various marine ecosystems. Until now, we have collected audio data from algal reefs, estuaries, seagrass beds,  coral reefs,  hydrothermal vents,  continental  shelves, and  deep sea  benthic environments. We also developed the Soundscape Viewer, which is an open toolbox of soundscape information retrieval (SIR), to enable efficient processing of a large amount of audio data.
The current framework of SIR integrates machine learning in source separation and event identification so that users can identify
various biotic and abiotic sounds in an unsupervised manner. In this talk, I will introduce the coral reef soundscapes off Okinawa, Japan, and the deep sea soundscapes off northeastern Taiwan to discuss the feasibility of soundscape based ecosystem monitoring.  Now, we are building a science platform of soundscapes that shares open tools and open data. By doing so, we hope managers and stakeholders can use soundscape in formation to monitor the trends of marine ecosystems and perform a data driven decision making in
conservation management.