{"url":"/dataset/dear","name":"DEAR","full_name":null,"description_markdown":"## Dataset Summary\r\n\r\nThe Deep Evaluation of Audio Representations (DEAR) dataset is a benchmark designed to assess general-purpose audio foundation models on properties critical for hearable devices. \r\nIt comprises **1,158** mono audio tracks (30 s each), spatially mixing proprietary anechoic speech monologues with high-quality everyday acoustic scene recordings from the HOA‑SSR library. \r\nDEAR enables controlled evaluation of:\r\n\r\n* **Context** (environment type: domestic, leisure, nature, professional, transport; indoor/outdoor; stationary/transient noise)\r\n* **Speech sources** (speech presence detection; speaker count)\r\n* **Acoustic properties** (direct-to-reverberant ratio DRR, reverberation time RT60, signal‑to‑noise ratio SNR)\r\n\r\nAll tracks are down‑mixed to a single channel at 44.1 kHz (32‑bit) and split into development and test sets with no overlap in speakers, backgrounds, or impulse responses.\r\n\r\n## Tasks\r\n\r\n| Task Group    | Task                                | Type        | Metric      |\r\n| ------------- | ----------------------------------- | ----------- | ----------- |\r\n| Context       | 5‑way environment classification    | Multi‑class | Matthews' $\\phi$ |\r\n|               | Indoor vs. outdoor                  | Binary      | Matthews' $\\phi$ |\r\n|               | Stationary vs. transient noise      | Binary      | Matthews' $\\phi$ |\r\n| Sources       | Speech presence (1 s segments)      | Binary      | Matthews' $\\phi$ |\r\n|               | Speaker count (1 s segments)        | Regression  | $R^2$          |\r\n| Acoustics     | DRR (1 s segments, 1 speaker)       | Regression  | $R^2$          |\r\n|               | RT60 (1 s segments, 1 speaker)      | Regression  | $R^2$          |\r\n|               | SNR (1 s segments, 1 speaker)       | Regression  | $R^2$          |\r\n| Retrospective | TUT2017 acoustic scene (15 classes) | Multi‑class | Matthews' $\\phi$ |\r\n|               | LibriCount speaker count (0–10)     | Regression  | $R^2$          |\r\n\r\n## Dataset Structure\r\n\r\n```\r\n├── data/\r\n│   ├── 00094903-4dbf-44a9-bf09-698fc361dbff.wav\r\n│   └── …\r\n├── development.csv\r\n└── test.csv\r\n```\r\n\r\n* **.wav files**: mono, 44.1 kHz, 32‑bit float\r\n* **.csv files**: meta-data for all tasks, linkable to wav files with `id`\r\n\r\n## Usage\r\n\r\nVisit the dedicated code repository: https://github.com/DEAR-dataset/code\r\n\r\n## Source Data\r\n\r\n* Speech monologues (proprietary anechoic recordings)\r\n* HOA‑SSR library ambisonics scenes (licensed via FORCE Technology)\r\n* Impulse responses for controlled reverberation\r\n\r\n## Citation\r\n\r\nIf you use DEAR in your research, please cite:\r\n\r\n```bibtex\r\n@inproceedings{\r\n  groeger2025dear,\r\n  author={Gröger, Fabian and Baumann, Pascal and Amruthalingam, Ludovic and Simon, Laurent and Giurda, Ruksana and Lionetti, Simone},\r\n  booktitle={ICASSP 2025 - 2025 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)}, \r\n  title={Evaluation of Deep Audio Representations for Hearables}, \r\n  year={2025},\r\n  doi={10.1109/ICASSP49660.2025.10887737}\r\n}\r\n```\r\n\r\nArXiv version: arxiv.org/abs/2502.06664","description_withheld":null,"homepage":"https://dear-dataset.github.io/","introduced_date":null,"introduced_date_note":null,"introduced_by":null,"license":{"name":"Creative Commons Attribution Non Commercial No Derivatives 4.0 International","url":null},"modalities":[{"name":"Audio","url":"/datasets/modality/audio"}],"tasks":[{"name":"Audio Classification","url":"/task/audio-classification","datasets_with_task":"/datasets/task/audio-classification"}],"languages":[],"variants":["DEAR"],"data_loaders":[{"repo":"https://github.com/DEAR-dataset/code","url":"https://github.com/DEAR-dataset/code","frameworks":["pytorch"]}],"num_papers_in_archive":0,"source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28"},"benchmarks":[],"papers_with_a_benchmark_row":[],"syntology_totals":{"read_at":"2026-09-24T18:15:14+00:00","papers_with_samples":0,"samples_harvested":0,"samples_ran":0,"samples_unverified":0,"pointer_only_for_licence":0,"papers_with_no_sample_that_ran":0,"note":"the per-paper counts above, summed; not a rate"},"papers_note":"The archive never published its papers-using-dataset list; these are papers with a leaderboard row on this dataset's benchmarks."}