README.md

metadata

license: other
task_categories:
  - text-retrieval
language:
  - en
  - zh
tags:
  - audio-text-retrieval
  - spoken-query-retrieval
  - mteb
pretty_name: SQuTR for MTEB
configs:
  - config_name: fiqa-corpus
    data_files:
      - split: test
        path: fiqa-corpus/test-*
  - config_name: fiqa-qrels
    data_files:
      - split: test
        path: fiqa-qrels/test-*
  - config_name: fiqa-queries-clean
    data_files:
      - split: test
        path: fiqa-queries-clean/test-*
  - config_name: fiqa-queries-snr-0
    data_files:
      - split: test
        path: fiqa-queries-snr-0/test-*
  - config_name: fiqa-queries-snr-10
    data_files:
      - split: test
        path: fiqa-queries-snr-10/test-*
  - config_name: fiqa-queries-snr-20
    data_files:
      - split: test
        path: fiqa-queries-snr-20/test-*
  - config_name: hotpotqa-corpus
    data_files:
      - split: test
        path: hotpotqa-corpus/test-*
  - config_name: hotpotqa-qrels
    data_files:
      - split: test
        path: hotpotqa-qrels/test-*
dataset_info:
  - config_name: fiqa-corpus
    features:
      - name: id
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: string
    splits:
      - name: test
        num_bytes: 45303212
        num_examples: 57638
    download_size: 27711317
    dataset_size: 45303212
  - config_name: fiqa-qrels
    features:
      - name: query-id
        dtype: string
      - name: corpus-id
        dtype: string
      - name: score
        dtype: int32
    splits:
      - name: test
        num_bytes: 37172
        num_examples: 1706
    download_size: 20162
    dataset_size: 37172
  - config_name: fiqa-queries-clean
    features:
      - name: id
        dtype: string
      - name: audio
        dtype: audio
      - name: transcript
        dtype: string
      - name: snr_db
        dtype: int32
      - name: noise_id
        dtype: string
    splits:
      - name: test
        num_bytes: 155683707
        num_examples: 648
    download_size: 155682018
    dataset_size: 155683707
  - config_name: fiqa-queries-snr-0
    features:
      - name: id
        dtype: string
      - name: audio
        dtype: audio
      - name: transcript
        dtype: string
      - name: snr_db
        dtype: int32
      - name: noise_id
        dtype: string
    splits:
      - name: test
        num_bytes: 103829506
        num_examples: 648
    download_size: 103823286
    dataset_size: 103829506
  - config_name: fiqa-queries-snr-10
    features:
      - name: id
        dtype: string
      - name: audio
        dtype: audio
      - name: transcript
        dtype: string
      - name: snr_db
        dtype: int32
      - name: noise_id
        dtype: string
    splits:
      - name: test
        num_bytes: 103829506
        num_examples: 648
    download_size: 103824055
    dataset_size: 103829506
  - config_name: fiqa-queries-snr-20
    features:
      - name: id
        dtype: string
      - name: audio
        dtype: audio
      - name: transcript
        dtype: string
      - name: snr_db
        dtype: int32
      - name: noise_id
        dtype: string
    splits:
      - name: test
        num_bytes: 103829506
        num_examples: 648
    download_size: 103823934
    dataset_size: 103829506
  - config_name: hotpotqa-corpus
    features:
      - name: id
        dtype: string
      - name: title
        dtype: string
      - name: text
        dtype: string
    splits:
      - name: test
        num_bytes: 1621286119
        num_examples: 5233329
    download_size: 981075180
    dataset_size: 1621286119
  - config_name: hotpotqa-qrels
    features:
      - name: query-id
        dtype: string
      - name: corpus-id
        dtype: string
      - name: score
        dtype: int32
    splits:
      - name: test
        num_bytes: 636264
        num_examples: 14810
    download_size: 263904
    dataset_size: 636264

SQuTR for MTEB

MTEB-ready packaging of SQuTR, a bilingual spoken-query-to-text retrieval benchmark.

It contains FiQA, HotpotQA, NQ, MedicalRetrieval, DuRetrieval, and T2Retrieval. Each collection includes clean, 20 dB, 10 dB, and 0 dB audio queries.

Query transcripts are kept for provenance but are not exposed to evaluated MTEB models.

SQuTR's generated data is CC BY-SA 4.0. The packaged corpora and qrels keep their original dataset licenses, including NQ's CC BY-NC-SA 3.0 terms. See the source dataset for details.

Source revision: 2f1b041e2e98e0d28ed68fbcf22126ef247eb719

Citation

@misc{li2026squtrrobustnessbenchmarkspoken,
  author = {Yuejie Li and Ke Yang and Yueying Hua and Berlin Chen and Jianhao Nie and Yueping He and Caixin Kang},
  title = {SQuTR: A Robustness Benchmark for Spoken Query to Text Retrieval under Acoustic Noise},
  year = {2026},
  eprint = {2602.12783},
  archivePrefix = {arXiv},
  primaryClass = {cs.IR},
  url = {https://arxiv.org/abs/2602.12783}
}