Datasets:
audio audioduration (s) 5.2 136 | start_time float32 0 3.26k | end_time float32 6 3.27k | duration float32 5.2 136 | year int32 2.02k 2.02k | hydrophone stringclasses 3
values |
|---|---|---|---|---|---|
1,440.400024 | 1,452 | 11.6 | 2,021 | channel_0 | |
1,704.800049 | 1,711.599976 | 6.8 | 2,021 | channel_0 | |
1,311.599976 | 1,317.199951 | 5.6 | 2,023 | channel_0 | |
2,609.600098 | 2,623.199951 | 13.6 | 2,023 | channel_0 | |
2,637.199951 | 2,642.800049 | 5.6 | 2,023 | channel_0 | |
2,670.800049 | 2,694.399902 | 23.6 | 2,023 | channel_0 | |
2,792.399902 | 2,801.199951 | 8.8 | 2,023 | channel_0 | |
2,805.199951 | 2,815.199951 | 10 | 2,023 | channel_0 | |
291.200012 | 297.200012 | 6 | 2,023 | channel_0 | |
314.399994 | 319.600006 | 5.2 | 2,023 | channel_0 | |
693.200012 | 698.799988 | 5.6 | 2,023 | channel_0 | |
258.399994 | 267.200012 | 8.8 | 2,023 | channel_1 | |
270.799988 | 276.799988 | 6 | 2,023 | channel_1 | |
280.799988 | 302.799988 | 22 | 2,023 | channel_1 | |
30.799999 | 36 | 5.2 | 2,023 | channel_1 | |
319.200012 | 336 | 16.799999 | 2,023 | channel_1 | |
340.399994 | 353.600006 | 13.2 | 2,023 | channel_1 | |
364.399994 | 386.399994 | 22 | 2,023 | channel_1 | |
410 | 432.399994 | 22.4 | 2,023 | channel_1 | |
467.600006 | 482 | 14.4 | 2,023 | channel_1 | |
500 | 507.600006 | 7.6 | 2,023 | channel_1 | |
548.799988 | 556.799988 | 8 | 2,023 | channel_1 | |
568.799988 | 574.799988 | 6 | 2,023 | channel_1 | |
580 | 587.599976 | 7.6 | 2,023 | channel_1 | |
626.400024 | 638 | 11.6 | 2,023 | channel_1 | |
663.200012 | 674 | 10.8 | 2,023 | channel_1 | |
686.400024 | 692 | 5.6 | 2,023 | channel_1 | |
695.599976 | 700.799988 | 5.2 | 2,023 | channel_1 | |
712.400024 | 731.599976 | 19.200001 | 2,023 | channel_1 | |
72 | 78.400002 | 6.4 | 2,023 | channel_1 | |
134.399994 | 141.600006 | 7.2 | 2,023 | channel_0 | |
20.4 | 34.400002 | 14 | 2,023 | channel_0 | |
2,674 | 2,684.399902 | 10.4 | 2,020 | channel_0 | |
3,172.399902 | 3,179.199951 | 6.8 | 2,020 | channel_0 | |
194 | 211.600006 | 17.6 | 2,021 | channel_0 | |
215.600006 | 224.399994 | 8.8 | 2,021 | channel_0 | |
228 | 237.199997 | 9.2 | 2,021 | channel_0 | |
251.199997 | 256.799988 | 5.6 | 2,021 | channel_0 | |
262.799988 | 268.799988 | 6 | 2,021 | channel_0 | |
318.399994 | 323.600006 | 5.2 | 2,021 | channel_0 | |
328 | 334.799988 | 6.8 | 2,021 | channel_0 | |
339.200012 | 350 | 10.8 | 2,021 | channel_0 | |
358 | 365.600006 | 7.6 | 2,021 | channel_0 | |
379.200012 | 399.200012 | 20 | 2,021 | channel_0 | |
465.200012 | 470.399994 | 5.2 | 2,021 | channel_0 | |
47.200001 | 52.799999 | 5.6 | 2,021 | channel_0 | |
619.599976 | 625.599976 | 6 | 2,021 | channel_0 | |
645.599976 | 658.799988 | 13.2 | 2,021 | channel_0 | |
713.599976 | 728.400024 | 14.8 | 2,021 | channel_0 | |
732 | 761.599976 | 29.6 | 2,021 | channel_0 | |
764.799988 | 772.400024 | 7.6 | 2,021 | channel_0 | |
778.799988 | 797.599976 | 18.799999 | 2,021 | channel_0 | |
801.599976 | 808.400024 | 6.8 | 2,021 | channel_0 | |
812.400024 | 830 | 17.6 | 2,021 | channel_0 | |
842.400024 | 865.200012 | 22.799999 | 2,021 | channel_0 | |
898 | 903.599976 | 5.6 | 2,021 | channel_0 | |
907.200012 | 917.599976 | 10.4 | 2,021 | channel_0 | |
921.200012 | 942.400024 | 21.200001 | 2,021 | channel_0 | |
946 | 973.599976 | 27.6 | 2,021 | channel_0 | |
1,125.199951 | 1,136.400024 | 11.2 | 2,019 | channel_0 | |
1,485.599976 | 1,495.199951 | 9.6 | 2,019 | channel_0 | |
1,718.400024 | 1,728 | 9.6 | 2,019 | channel_0 | |
1,745.599976 | 1,752.800049 | 7.2 | 2,019 | channel_0 | |
657.599976 | 663.599976 | 6 | 2,019 | channel_0 | |
824.400024 | 831.599976 | 7.2 | 2,019 | channel_0 | |
894.799988 | 900.799988 | 6 | 2,019 | channel_0 | |
1,034 | 1,039.199951 | 5.2 | 2,023 | channel_1 | |
1,061.199951 | 1,087.199951 | 26 | 2,023 | channel_1 | |
1,099.199951 | 1,105.599976 | 6.4 | 2,023 | channel_1 | |
1,259.599976 | 1,264.800049 | 5.2 | 2,023 | channel_1 | |
1,417.199951 | 1,422.800049 | 5.6 | 2,023 | channel_1 | |
150 | 157.199997 | 7.2 | 2,023 | channel_1 | |
160.399994 | 167.199997 | 6.8 | 2,023 | channel_1 | |
1,660.400024 | 1,667.199951 | 6.8 | 2,023 | channel_1 | |
230.399994 | 236.800003 | 6.4 | 2,023 | channel_1 | |
526.400024 | 537.200012 | 10.8 | 2,023 | channel_1 | |
640 | 659.599976 | 19.6 | 2,023 | channel_1 | |
673.200012 | 680.400024 | 7.2 | 2,023 | channel_1 | |
730 | 737.599976 | 7.6 | 2,023 | channel_1 | |
749.200012 | 754.400024 | 5.2 | 2,023 | channel_1 | |
758.799988 | 768.799988 | 10 | 2,023 | channel_1 | |
79.199997 | 86 | 6.8 | 2,023 | channel_1 | |
837.200012 | 843.200012 | 6 | 2,023 | channel_1 | |
860.799988 | 869.200012 | 8.4 | 2,023 | channel_1 | |
909.200012 | 919.599976 | 10.4 | 2,023 | channel_1 | |
925.599976 | 940.400024 | 14.8 | 2,023 | channel_1 | |
954.799988 | 995.599976 | 40.799999 | 2,023 | channel_1 | |
998.799988 | 1,005.200012 | 6.4 | 2,023 | channel_1 | |
100 | 105.199997 | 5.2 | 2,023 | channel_0 | |
0 | 6 | 6 | 2,020 | channel_0 | |
1,265.199951 | 1,272 | 6.8 | 2,020 | channel_0 | |
1,590 | 1,596.400024 | 6.4 | 2,020 | channel_0 | |
190.800003 | 214.800003 | 24 | 2,020 | channel_0 | |
60.799999 | 67.599998 | 6.8 | 2,020 | channel_0 | |
1,144.400024 | 1,152 | 7.6 | 2,020 | channel_0 | |
1,174 | 1,179.599976 | 5.6 | 2,020 | channel_0 | |
1,192 | 1,212.400024 | 20.4 | 2,020 | channel_0 | |
1,799.199951 | 1,808.400024 | 9.2 | 2,020 | channel_0 | |
275.600006 | 283.200012 | 7.6 | 2,023 | channel_0 | |
1,785.599976 | 1,795.199951 | 9.6 | 2,019 | channel_0 |
OpenWhistle Pretraining Dataset
dolphinteam/OpenWhistle-Pretraining is the public unlabeled audio
dataset used for OpenWhistle pretraining. It contains 96 kHz dolphin acoustic
segments with timing and recording metadata, but no whistle/noise labels.
The default config is the complete pretraining dataset, with train and
validation splits.
Dataset contents
- Hugging Face repo:
dolphinteam/OpenWhistle-Pretraining - Public columns:
audio,start_time,end_time,duration,year,hydrophone - Sampling rate: 96 kHz, mono audio
- Labels: none; this dataset is intended for unsupervised or self-supervised pretraining
Dataset splits
| Split | Rows | Duration (s) | Duration (h) |
|---|---|---|---|
train |
28,410 | 367,792.80 | 102.165 |
validation |
3,370 | 43,629.20 | 12.119 |
| Total | 31,780 | 411,422.00 | 114.284 |
Coverage by year
| Split | 2019 | 2020 | 2021 | 2023 | 2024 |
|---|---|---|---|---|---|
train |
920 | 1,420 | 14,376 | 9,885 | 1,809 |
validation |
119 | 179 | 1,820 | 1,055 | 197 |
| Total | 1,039 | 1,599 | 16,196 | 10,940 | 2,006 |
Coverage by hydrophone
| Split | channel_0 | channel_1 | channel_2 |
|---|---|---|---|
train |
20,808 | 7,130 | 472 |
validation |
2,599 | 741 | 30 |
| Total | 23,407 | 7,871 | 502 |
Coverage by year and hydrophone
| Split | 2019 ch0 | 2020 ch0 | 2021 ch0 | 2023 ch0 | 2023 ch1 | 2023 ch2 | 2024 ch0 | 2024 ch1 |
|---|---|---|---|---|---|---|---|---|
train |
920 | 1,420 | 14,376 | 3,982 | 5,431 | 472 | 110 | 1,699 |
validation |
119 | 179 | 1,820 | 464 | 561 | 30 | 17 | 180 |
| Total | 1,039 | 1,599 | 16,196 | 4,446 | 5,992 | 502 | 127 | 1,879 |
Loading the data
from datasets import load_dataset
dataset = load_dataset("dolphinteam/OpenWhistle-Pretraining")
The full dataset is about 79 GB of parquet files. To inspect a few examples without downloading everything, stream it:
from datasets import load_dataset
stream = load_dataset("dolphinteam/OpenWhistle-Pretraining", split="validation", streaming=True)
example = next(iter(stream))
Earlier revisions
Earlier revisions of this repository also contained a 480-example
review-sample config for inspection. It remains available at revision
e0fbe77305944272ce24446293fd98a1b8c6f04d:
from datasets import load_dataset
review = load_dataset(
"dolphinteam/OpenWhistle-Pretraining",
"review-sample",
revision="e0fbe77305944272ce24446293fd98a1b8c6f04d",
)
License
This dataset is licensed under Creative Commons Attribution 4.0 International (CC BY 4.0), as recorded in its Croissant/RAI metadata.
Paper, collection, and authors
- Paper: OpenWhistle: A Large-Scale Longitudinal Dataset and Benchmark of Bottlenose Dolphin Vocalizations
- Hugging Face collection: NeurIPS 2026 OpenWhistle
- This repository: https://huggingface.co/datasets/dolphinteam/OpenWhistle-Pretraining
Authors: Faadil Mustun, Chiara Semenzin, Roberto Dessi, Pablo Robin Guerrero, Pierre Orhan, Alexis Emanuelli, Emanuele Rossi, Yair Lakretz, Gonzalo de Polavieja, and German Sumbre.
Citation
@misc{mustun2026openwhistle,
title = {OpenWhistle: A Large-Scale Longitudinal Dataset and Benchmark of Bottlenose Dolphin Vocalizations},
author = {Mustun, Faadil and Semenzin, Chiara and Dessi, Roberto and Robin Guerrero, Pablo and Orhan, Pierre and Emanuelli, Alexis and Rossi, Emanuele and Lakretz, Yair and de Polavieja, Gonzalo and Sumbre, German},
year = {2026},
eprint = {2609.34839},
archivePrefix = {arXiv},
primaryClass = {cs.CL},
url = {https://arxiv.org/abs/2609.34839}
}
- Downloads last month
- 2,349