--- dataset_info: - config_name: all features: - name: audio dtype: audio: decode: false - name: label dtype: class_label: names: '0': NSW_3 '1': NSW_2 '2': NSW_1 '3': SW_Dana '4': SW_Luna '5': SW_Nana '6': SW_Neo '7': SW_Nikita '8': SW_Shy '9': SW_Yosefa - name: name dtype: string - name: onset dtype: float32 - name: offset dtype: float32 - name: duration dtype: float32 - name: recording_duration dtype: float32 - name: whistle_type dtype: int64 - name: whistle_name dtype: string - name: f0_time sequence: float32 - name: f0_hz sequence: float32 - name: f0_conf sequence: float32 - name: f0_ok dtype: bool - name: f0_bad_reason dtype: string - name: f0_spectrogram dtype: image - name: snr_db dtype: float64 splits: - name: train num_examples: 5848 - name: validation num_examples: 1253 - name: test num_examples: 1253 - config_name: all-review-sample features: - name: audio dtype: audio: decode: false - name: label dtype: class_label: names: '0': NSW_3 '1': NSW_2 '2': NSW_1 '3': SW_Dana '4': SW_Luna '5': SW_Nana '6': SW_Neo '7': SW_Nikita '8': SW_Shy '9': SW_Yosefa - name: name dtype: string - name: onset dtype: float32 - name: offset dtype: float32 - name: duration dtype: float32 - name: recording_duration dtype: float32 - name: whistle_type dtype: int64 - name: whistle_name dtype: string - name: f0_time sequence: float32 - name: f0_hz sequence: float32 - name: f0_conf sequence: float32 - name: f0_ok dtype: bool - name: f0_bad_reason dtype: string - name: f0_spectrogram dtype: image - name: snr_db dtype: float64 splits: - name: train num_examples: 336 - name: validation num_examples: 72 - name: test num_examples: 72 - config_name: balanced features: - name: audio dtype: audio: decode: false - name: label dtype: class_label: names: '0': NSW_1 '1': SW_Luna '2': SW_Nana '3': SW_Neo '4': SW_Nikita '5': SW_Yosefa - name: name dtype: string - name: onset dtype: float32 - name: offset dtype: float32 - name: duration dtype: float32 - name: recording_duration dtype: float32 - name: whistle_type dtype: int64 - name: whistle_name dtype: string - name: f0_time sequence: float32 - name: f0_hz sequence: float32 - name: f0_conf sequence: float32 - name: f0_ok dtype: bool - name: f0_bad_reason dtype: string - name: f0_spectrogram dtype: image splits: - name: train num_examples: 2100 - name: validation num_examples: 450 - name: test num_examples: 450 - config_name: balanced-review-sample features: - name: audio dtype: audio: decode: false - name: label dtype: class_label: names: '0': NSW_1 '1': SW_Luna '2': SW_Nana '3': SW_Neo '4': SW_Nikita '5': SW_Yosefa - name: name dtype: string - name: onset dtype: float32 - name: offset dtype: float32 - name: duration dtype: float32 - name: recording_duration dtype: float32 - name: whistle_type dtype: int64 - name: whistle_name dtype: string - name: f0_time sequence: float32 - name: f0_hz sequence: float32 - name: f0_conf sequence: float32 - name: f0_ok dtype: bool - name: f0_bad_reason dtype: string - name: f0_spectrogram dtype: image splits: - name: train num_examples: 336 - name: validation num_examples: 72 - name: test num_examples: 72 - config_name: unbalanced features: - name: audio dtype: audio: decode: false - name: label dtype: class_label: names: '0': NSW_3 '1': NSW_2 '2': NSW_1 '3': SW_Dana '4': SW_Luna '5': SW_Nana '6': SW_Neo '7': SW_Nikita '8': SW_Shy '9': SW_Yosefa - name: name dtype: string - name: onset dtype: float32 - name: offset dtype: float32 - name: duration dtype: float32 - name: recording_duration dtype: float32 - name: whistle_type dtype: int64 - name: whistle_name dtype: string - name: f0_time sequence: float32 - name: f0_hz sequence: float32 - name: f0_conf sequence: float32 - name: f0_ok dtype: bool - name: f0_bad_reason dtype: string - name: f0_spectrogram dtype: image splits: - name: train num_examples: 2442 - name: validation num_examples: 523 - name: test num_examples: 523 - config_name: unbalanced-review-sample features: - name: audio dtype: audio: decode: false - name: label dtype: class_label: names: '0': NSW_3 '1': NSW_2 '2': NSW_1 '3': SW_Dana '4': SW_Luna '5': SW_Nana '6': SW_Neo '7': SW_Nikita '8': SW_Shy '9': SW_Yosefa - name: name dtype: string - name: onset dtype: float32 - name: offset dtype: float32 - name: duration dtype: float32 - name: recording_duration dtype: float32 - name: whistle_type dtype: int64 - name: whistle_name dtype: string - name: f0_time sequence: float32 - name: f0_hz sequence: float32 - name: f0_conf sequence: float32 - name: f0_ok dtype: bool - name: f0_bad_reason dtype: string - name: f0_spectrogram dtype: image splits: - name: train num_examples: 336 - name: validation num_examples: 72 - name: test num_examples: 72 task_categories: - audio-classification tags: - dolphin - bioacoustics - whistle-classification - audio - f0 - spectrogram configs: - config_name: all data_files: - split: train path: all/train-* - split: validation path: all/validation-* - split: test path: all/test-* - config_name: all-review-sample data_files: - split: train path: all-review-sample/train-* - split: validation path: all-review-sample/validation-* - split: test path: all-review-sample/test-* - config_name: balanced data_files: - split: train path: balanced/train-* - split: validation path: balanced/validation-* - split: test path: balanced/test-* - config_name: balanced-review-sample data_files: - split: train path: balanced-review-sample/train-* - split: validation path: balanced-review-sample/validation-* - split: test path: balanced-review-sample/test-* - config_name: unbalanced data_files: - split: train path: unbalanced/train-* - split: validation path: unbalanced/validation-* - split: test path: unbalanced/test-* - config_name: unbalanced-review-sample data_files: - split: train path: unbalanced-review-sample/train-* - split: validation path: unbalanced-review-sample/validation-* - split: test path: unbalanced-review-sample/test-* --- # OpenWhistle Classification Finetuning Dataset `OpenWhistleNeurIPS26/OpenWhistle-Classification-Finetuning` is the public classification finetuning dataset used for dolphin whistle identity classification. It contains short whistle clips, whistle-level metadata, fundamental-frequency tracks, rendered F0 spectrograms, and integer class labels. The main reviewer-facing subset is the balanced `balanced` config. It contains six classes: - `NSW_1` (`label=0`) - `SW_Luna` (`label=1`) - `SW_Nana` (`label=2`) - `SW_Neo` (`label=3`) - `SW_Nikita` (`label=4`) - `SW_Yosefa` (`label=5`) The full dataset is split by recording session, so no session appears in more than one of `train`, `validation`, or `test`. Smaller deterministic review configs are also provided so reviewers can inspect representative examples quickly without downloading the complete data first. ## Dataset Contents - Hugging Face repo: `OpenWhistleNeurIPS26/OpenWhistle-Classification-Finetuning` - Main balanced config: `balanced` - Reviewer convenience config: `balanced-review-sample` - Public columns common to all configs: `audio`, `label`, `name`, `onset`, `offset`, `duration`, `recording_duration`, `whistle_type`, `whistle_name`, `f0_time`, `f0_hz`, `f0_conf`, `f0_ok`, `f0_bad_reason`, `f0_spectrogram` - The `all` and `all-review-sample` configs additionally include `snr_db`, the estimated clip-level signal-to-noise ratio in dB. ## Balanced Dataset Splits | Split | Rows | NSW_1 | SW_Luna | SW_Nana | SW_Neo | SW_Nikita | SW_Yosefa | Sessions | | --- | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | | `train` | 2,100 | 350 | 350 | 350 | 350 | 350 | 350 | 161 | | `validation` | 450 | 75 | 75 | 75 | 75 | 75 | 75 | 54 | | `test` | 450 | 75 | 75 | 75 | 75 | 75 | 75 | 37 | | **Total** | **3,000** | **500** | **500** | **500** | **500** | **500** | **500** | **252** | The split assignment was generated with seed `42` and exact class balancing. Session leakage checks found no overlap between any pair of splits. ## Available Subsets The repository provides three full classification subsets and their smaller review counterparts: | Subset/config | Rows | Classes | Sessions | Purpose | | --- | ---: | ---: | ---: | --- | | `balanced` | 3,000 | 6 | 252 | Main balanced six-class finetuning dataset | | `unbalanced` | 3,488 | 10 | 258 | Ten-class finetuning dataset with capped rare classes | | `all` | 8,354 | 10 | 261 | Ten-class dataset preserving the full available class distribution | | `balanced-review-sample` | 480 | 6 | Same source split design | Small reviewer sample from `balanced` | | `unbalanced-review-sample` | 480 | 10 | Same source split design | Small reviewer sample from `unbalanced` | | `all-review-sample` | 480 | 10 | Same source split design | Small reviewer sample from `all` | The ten-class subsets use the following labels: - `NSW_3` - `NSW_2` - `NSW_1` - `SW_Dana` - `SW_Luna` - `SW_Nana` - `SW_Neo` - `SW_Nikita` - `SW_Shy` - `SW_Yosefa` The `balanced` subset keeps the six classes listed above and is the recommended starting point for reviewers and model finetuning. The `unbalanced` and `all` subsets expose the broader ten-class label space for additional analysis. ## Review Samples Review samples are small deterministic subsets of the same public dataset. They were created only to make review and manual inspection easier. They are not a replacement for the full configs used for model development or reporting. ### How The Review Samples Were Created All review samples were built after the session-disjoint train/validation/test splits were finalized. The review-sample scripts preserve the original split assignment: reviewer training examples come only from the original `train` split, reviewer validation examples only from `validation`, and reviewer test examples only from `test`. For `balanced-review-sample`, rows were sampled separately within each split and class. Each class group was shuffled deterministically with `numpy.default_rng(seed + split_index)` using seed `42`, then capped at 56 rows per class for `train` and 12 rows per class for both `validation` and `test`. This keeps the same 70/15/15 split ratio as the full `balanced` config while keeping every class equally represented. For `unbalanced-review-sample` and `all-review-sample`, the same deterministic shuffle was used, but the target rows were allocated proportionally to the source class distribution inside each split. This preserves the class imbalance of the larger source configs while keeping the review download small. ### Review Sample Sizes | Config | Source config | Strategy | Train | Validation | Test | Total | | --- | --- | --- | ---: | ---: | ---: | ---: | | `balanced-review-sample` | `balanced` | Equal rows per class within each split | 336 | 72 | 72 | 480 | | `unbalanced-review-sample` | `unbalanced` | Proportional class distribution within each split | 336 | 72 | 72 | 480 | | `all-review-sample` | `all` | Proportional class distribution within each split | 336 | 72 | 72 | 480 | The reviewer-facing sample for the main balanced dataset is `balanced-review-sample`. The other review samples are included so each full subset has a matching small inspection subset. ## Loading The Data ```python from datasets import load_dataset full = load_dataset( "OpenWhistleNeurIPS26/OpenWhistle-Classification-Finetuning", "balanced", ) review = load_dataset( "OpenWhistleNeurIPS26/OpenWhistle-Classification-Finetuning", "balanced-review-sample", ) ``` Optional broader configs can be loaded by passing `"unbalanced"` or `"all"` as the second `load_dataset` argument. Their corresponding review configs are `"unbalanced-review-sample"` and `"all-review-sample"`.