Upload README.md with huggingface_hub

README.md

@@ -2,11 +2,20 @@
 license: cc-by-4.0
 task_categories:
 - audio-to-audio
-pretty_name: Dominica Sperm Whale Project
+tags:
+- bioacoustics
+- sperm-whale
+- marine-biology
+- arxiv:2512.02206
+- Project-CETI/wham
+pretty_name: Dominica Sperm Whale Project (DSWP)
 size_categories:
 - 1K<n<10K
 ---
 # Dominica Sperm Whale Project (DSWP) dataset
+[![Code](https://img.shields.io/badge/GitHub-Project--CETI%2Fwham-blue?logo=github)](https://github.com/Project-CETI/wham)
+[![Paper](https://img.shields.io/badge/arXiv-2512.02206-b31b1b.svg)](https://arxiv.org/abs/2512.02206)
+
 This dataset contains 1,501 sperm whale codas (~45 minutes of audio) recorded off the coast of Dominica. Each audio file contains at least one coda extracted from longer recordings collected by the [Dominica Sperm Whale Project](https://www.thespermwhaleproject.org/) using a combination of far-field boat-based hydrophones and animal-borne acoustic tags.
 
 This dataset accompanies our paper:
@@ -16,9 +25,9 @@ General background about codas can be found in Appendix B therein.
 
 ### Data Collection
 Codas were collected between 2005 and 2018 during field seasons in a ~2000 km² area off the coast of Dominica. Recordings were obtained using several passive, non-invasive recording systems over the years, including towed hydrophones and animal-borne tags. Specifically:
-- A Fostex VF-160 multitrack recorder (44.1kHz sampling rate) and a custom built towed hydrophone (Benthos AQ-4 elements, frequency response: 0.1--30kHz) with a filter box with high-pass filters up to 1 kHz resulting in a recording chain with a flat frequency response across a minimum of 2--20kHz.
-- A Zoom H4 portable field recorder (48kHz sampling rate) and a Cetacean Research Technology C55 hydrophone (frequency response: 0.02--44kHz) and no filters.
-- A custom-built towed hydrophone (Benthos AQ-4 elements, frequency response: 0.1--30kHz) with a filter box with high-pass filters up to 1 kHz resulting in a recording chain with a flat frequency response across a minimum of 2--20 kHz. This was connected to a computer based recording system as a part of the International Fund for Animal Welfare’s (IFAW) LOGGER software package (48kHz sampling rate) or PAMGUARD (minimum 48 kHz sampling rate). In addition, recordings were also made through the deployment of animal-borne sound and movement tags (DTag generation 3; _Johnson & Tyack 2003_).
+- Towed Hydrophones: Benthos AQ-4 elements, 0.1–30 kHz response, with high-pass filter boxes (flat response 2--20 kHz). These were connected to either a Fostex VF-160 multitrack recorder (44.1 kHz) or a computer-based system running IFAW’s LOGGER or PAMGUARD (min. 48 kHz).
+- Portable Recorders: A Zoom H4 field recorder (48 kHz) with a Cetacean Research Technology C55 hydrophone (0.02--44 kHz).
+- Whale-borne tag: DTag generation 3 sound and movement tags (_Johnson & Tyack 2003_).
 
 ### Intended use and limitations
 This dataset could be used for a variety of audio tasks, including e.g. analysis of sperm whale coda structure, machine learning for audio representation, or comparative studies of animal communication signals. Limitations are that codas are presented as isolated snippets without behavioral context, there is no per-file metadata (e.g. which recording system was used), and that all recordings came from the same vocal clan and geographic region.