What is the purpose of this project?

This is a long-term project that aims to identify, categorize, and classify dolphin phonations. We need the help of citizen scientists to train our deep learning models to do these tasks automatically so they can work in real time with passive acoustic data. The ultimate goal is to create a system for rapidly identifying and classifying dolphin phonations so we can start searching for underlying order and patterns that will help us to try and understand dolphin communication system.

Where and when were these recordings made?

All recordings were made in the presence of wild bottlenose dolphins in the Tampa Bay area and the Mississippi Sound area. The recordings were made during daytime, between 8 am and 4 pm.

I cannot distinguish snapping shrimp from dolphin clicks; what should I do?

It is a common problem and there are two helpful tips that could help. First, dolphin echolocation, click trains will always have high frequency component, so look for high frequency components. While some snapping shrimp sounds could extend to ultrasound range, the majority do not. Second, dolphin clicks have a certain order or pattern present, but snapping shrimp clicks are largely random, as many shrimps are clicking at the same time.

Why do some sounds look so faint?

In many cases, the dolphin(s) were far away from the recording system, so were were only able to capture the partial phonations that they made. In other cases, noises like boat noise or snapping shrimp could actually mask dolphin phonations.

Why are we asked to identify sounds other than dolphin phonations?

Understanding dolphin habitat soundscape is very important because it could potentially affect their communication system and patterns of communication. As you will hear, both habitats have very rich soundscapes, so it is important to identify all sounds of interest so our deep learning models can learn to identify these other sounds too.

What is the final count of subjects?

This is a long-term and we will be adding datasets as we go along. We estimate that the total number of subjects will be around 20000-25000 files.

Are there breathing sounds in the files?

No; just like manatees, dolphin breathe above water, so there will be no breathing sounds in the sound files.

I am not sure what category a sound belongs to; what should I do?

We recommend, if unsure, go conservative. Each file will be examined by at least 15 citizen scientists and we will see files with a lot of disagreements, so we can take a closer look at them.