zzDFTBA's picture
|
download
raw
1.47 kB
metadata
license: cc-by-4.0
language:
  - zh
pretty_name: ScreenASR-Bench
task_categories:
  - automatic-speech-recognition
tags:
  - audio
  - image
  - multimodal
  - benchmark
configs:
  - config_name: default
    data_files:
      - split: test
        path: data/test-*.parquet

ScreenASR-Bench

Data

Item Value
Split test
Cases 2,002
Audio clips 2,002
Keyframes 2,469
Languages Chinese

Structure

Field Type Description
caseid string Unique case identifier
ref string Reference transcription
target string Target text in TN form
level string Difficulty level: L1, L2, or L3
audio audio Audio clip
keyframes list[image] Keyframes associated with the case
frame_captions list[string] Captions aligned with keyframes

License

This benchmark is released under CC BY-NC 4.0 for academic research only. Commercial use is prohibited.

Citation

If you find ScreenASR-Bench useful in your research, please consider citing our paper:

@misc{fu2026screenasrbenchtieredbenchmarkspeech,
  title={ScreenASR-Bench: A Tiered Benchmark for Speech Recognition with Visual Context},
  author={Mingwei Fu and Zheshu Song and Zhikai Zhou and Shen Huang and Shidong Shang and Mengxiao Zhu},
  year={2026},
  eprint={2610.16260},
  archivePrefix={arXiv},
  primaryClass={cs.AI},
  url={https://arxiv.org/abs/2610.16260}
}

Xet Storage Details

Size:
1.47 kB
·
Xet hash:
67b0ad676ce070265d40b678b1e35376e2cc612772e53ced0050b5b267e46d2b

Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.