Humans & machines listening together to save the whales.

Orcasound is an open source project that makes it easy to listen live for animals that make sounds, like the endangered orcas in Washington State (aka Southern Resident Killer Whales, or SRKWs). We use underwater microphones (hydrophones) deployed in key habitats to listen for the calls, whistles, and clicks that SRKWs make almost continuously -- intense signals that can be heard 10 km away, often before the orcas are visible.

Since a Kickstarter in mid-2017, key members of our community have built a web app that is connected to 3 hydrophones monitoring the SRKW habitat 24/7. Version 2 of the app, now in beta-testing, lets citizen scientists indicate when they detect orca sounds in the live audio data streams, or other sounds that might harm the orcas, like sonar or noise from ships. The web-app repository has multiple feature branches. Users of version 2 can also receive a notification when orcas are detected, a feature which makes Orcasound a new conservation tool. While users enjoy the live “concert” they are offered a timely call to action -- often aimed at reducing the underwater noise pollution near Seattle which impedes the orcas’ ability to communicate & echolocate scarce salmon.

Orcasound also maintains repositories with other open source tools that support the web app, as well as bioacoustic analysis in general, and killer whale recovery:

  • orcanode holds Python code and shell scripts that generate the live audio stream in lossy HLS or DASH format to optimize audio playback within modern browsers and lossless FLAC format for scientific analysis.
  • orcadata houses multiple team efforts to develop machine learning algorithms and data processing tools for bioacoustics and underwater noise pollution.
  • orcabooth shares Orcasound recordings with educational/outreach organizations.
  • orcamap visualizes acoustic detection and classification data spatially.
lightbulb_outline View ideas list

Technologies

  • elixir
  • react
  • javascript
  • python
  • raspberry pi

Topics

comment IRC Channel
email Mailing list
mail_outline Contact email

Orcasound 2020 Projects

  • Kunal Mehta
    Active Listening & Learning of Orca Sounds
    The aim of the project is to create an Active Deep Learning and Machine Learning model that would predict the class of the calls depending on the...
  • Diego R. Saltijeral
    OrcaAL | Orca detection Active Learning tool
    Underwater noise pollution impedes the orcas’ ability to communicate and echolocate scarce salmon. To be able to detect orca calls, a deep learning...
close

2020