24
$\begingroup$

I would like to automatize the analysis of marine soundscapes by training a neural network that can detect and identify marine species. However, I need a large amount of training data to build such model.

I have searched for collections / libraries of marine species' vocalization or noise (something similar to xeno canto but for marine species) but I did not found anything that I can use. For instance this paper describes the need of gathering existing libraries but the idea hasn't been concertize yet.

Any help would be appreciated!

Improve this question
$\endgroup$

6 Answers 6

Reset to default
18
$\begingroup$

There are some resources available, but often hampered by the sheer size of these acoustic datasets which makes it difficult to keep them online. http://www.mobysound.org/ has some examples. Also check out the datasets from the DCLDE workshop series. A link to the latest of these is https://www.soest.hawaii.edu/ore/dclde/. From there you can follow links to previous meetings. Each workshop provided an annotated dataset for download.

Improve this answer
$\endgroup$
17
$\begingroup$

For fish sounds, there is also a relatively new and growing site, https://fishsounds.net, which hosts some downloadable recordings along with spectrograms, and citations for where they were described. For some species, links to other libraries are given.

Best of luck with your project!

Improve this answer
$\endgroup$
16
$\begingroup$

Nothing as comprehensive as xenocanto for the marine realm. At Fishbase https://www.fishbase.se/search.php, you can filter for sound files ("Information by Topic"), and there's a pretty decent list of results. The Watkins Marine Mammal Sound Database has a collection as well: https://cis.whoi.edu/science/B/whalesounds/index.cfm. At the Macaulay Library https://www.macaulaylibrary.org/ you can filter by sounds, but you have to request files from them directly and my students have had varied success in response to their requests.

Improve this answer
$\endgroup$
14
$\begingroup$

Dosits' library might also be of help. Here's the link to their audio gallery https://dosits.org/galleries/audio-gallery/

All the best of luck, sounds like a great project!

Improve this answer
$\endgroup$
12
$\begingroup$

Cornell University's Macaulay Library has a very comprehensive and curated collection of avain and marine mammal sounds. They are also rated by quality if you want to test your neural network. However, you need to submit an official request to gain access to the sound files, and they are pretty strict on the sounds being used for educational or testing purposes.

For a large collection of marine soundscapes in different environments, DOSITS is a great place to start.

Improve this answer
$\endgroup$
3
  • $\begingroup$ Hi @etgriffiths. I've edited the hyperlinks to follow Markdown syntax. However the Macaulay library link is not valid - please update/fix? $\endgroup$
    – Thejasvi
    Commented Jun 27, 2022 at 12:50
  • $\begingroup$ As it’s currently written, your answer is unclear. Please edit to add additional details that will help others understand how this addresses the question asked. You can find more information on how to write good answers in the help center. $\endgroup$
    – Community Bot
    Commented Jun 27, 2022 at 12:50
  • $\begingroup$ Apologies. I wrote the comment on my phone and the links didn't translate well. :P $\endgroup$
    – etgriffiths
    Commented Jun 27, 2022 at 14:37
8
$\begingroup$

This site has a compilation of publicly available bioacoustic datasets & repo's (not necessarily specific to marine, but has some marine datasets included). https://bioacousticsdatasets.weebly.com/index.html#datasets

There are also some listed on LILA BC's acoustics page here: https://lila.science/otherdatasets#bioacoustics

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.