On Collaborating With The Watkins Marine Mammal Sound Database:

[oliver-adams-b]

Hey all! I've been on the hunt for a large selection of vocalization data, and I've since stumbled onto the Watkins Database. I'm looking to source as much data from them as possible, and so connecting with them and green-lighting this is important. Just want to pose a few questions here: does anyone already have a connection to/with the Watkins Database? On their website it is stated that the data is open to personal and academic use, and I would like to know what ESP's standards/best practices are for shaping a relationships with researchers?

[radekosmulski]

Hey Oliver!

First of all, really appreciate your efforts on this! I gave this a bit more thought and here is my current thinking. I think between the licence under which they share their files and willingness to be a good citizen of the research community, you should be okay to scrape parts of the Watkins library. I think 'parts' would be the key phrase here. I don't think it would be appreciated if we were to copy the library in its entirety (even if it was within the licence) but if you were to created a curated dataset from a portion of the library and make it accessible through the untar_data interface of fastai v2, that would work really well and should be within the boundaries of good citizenship of the research community, I feel.

I feel this is the correct and safe way to approach this. But if anyone else would feel differently, this thread would be a great place to discuss any concerns! 😊 Thx for kicking off the discussion here Oliver!