The Orchive

There are stable resident populations of Orcinus orca in the northwest Pacific Ocean, and some of these populations are found near Hanson Island, off the north tip of Vancouver Island in Canada. Orcalab is a research station that has been recording audio of these Orca populations since 1972. They have amassed a huge archive of more than 20,000 hours of audio recordings collected via a permanent installation of underwater hydrophones.

In collaboration with OrcaLab, our lab at UVIC has digitized all these audio cassettes. The Orchive is a website that we have developed to make these recordings accessible to researchers.

To make these datasets even more accessible for researchers in the fields of Music Information Retrieval, Signal Processing and Machine Learning, we here present all the raw audio files for your use. If you have any questions, please email me at sness [at]

We also have a catalog of calls of the Northern Resident Killer Whales, collected by OrcaLab and marked up by hand by Helena Symonds. The original call catalog can be downloaded in gzipped tar format : orchive-call-catalog.tgz

In order to facilitate use of this call catalog by computer systems we have trimmed off the leading and trailing silence, these clips can be downloaded from: orchive-call-catalog-xsilence.tgz

A team of 10 scientists and experts in bioacoustics have made over 17,000 annotations of the Orchive. These annotations can be accessed using the main Orchive but can also be downloaded here : orchive-extract.tgz

In order to further make this set of recordings useful for researchers in machine learning, I have made a collection of 287 clips from this archive that are of high quality and are of a smaller set of calls. This can be downloaded here: orchive-audible.tgz

While you can view all the labbooks using the raw pdf files.

If you do use these recordings, please cite the following Ph.D. thesis:

The Orchive : A system for semi-automatic annotation and analysis of a large collection of bioacoustic recordings

  title={The Orchive: A system for semi-automatic annotation and analysis of a large collection of bioacoustic recordings},
  author={Ness, Steven},