Datasets:

Modalities:
Audio
Text
Formats:
parquet
Languages:
English
ArXiv:
Libraries:
Datasets
pandas
License:
voicebench / README.md
mattymchen's picture
Update README.md
aaa46d4 verified
metadata
language:
  - en
dataset_info:
  - config_name: advbench
    features:
      - name: audio
        dtype: audio
      - name: prompt
        dtype: string
    splits:
      - name: test
        num_bytes: 80626668
        num_examples: 520
    download_size: 74970333
    dataset_size: 80626668
  - config_name: alpacaeval
    features:
      - name: audio
        dtype: audio
      - name: prompt
        dtype: string
    splits:
      - name: test
        num_bytes: 36123845
        num_examples: 199
    download_size: 33542907
    dataset_size: 36123845
  - config_name: alpacaeval_full
    features:
      - name: audio
        dtype: audio
      - name: prompt
        dtype: string
    splits:
      - name: test
        num_bytes: 140044143
        num_examples: 636
    download_size: 130048839
    dataset_size: 140044143
  - config_name: alpacaeval_speaker
    features:
      - name: audio
        dtype: audio
      - name: prompt
        dtype: string
    splits:
      - name: en_AU_Wavenet_A_1.0_0.0_0.0
        num_bytes: 141127023
        num_examples: 636
      - name: en_AU_Wavenet_B_1.0_0.0_0.0
        num_bytes: 120085743
        num_examples: 636
      - name: en_IN_Wavenet_A_1.0_0.0_0.0
        num_bytes: 135913071
        num_examples: 636
      - name: en_IN_Wavenet_B_1.0_0.0_0.0
        num_bytes: 122491119
        num_examples: 636
      - name: en_GB_Wavenet_A_1.0_0.0_0.0
        num_bytes: 136140015
        num_examples: 636
      - name: en_GB_Wavenet_B_1.0_0.0_0.0
        num_bytes: 130439919
        num_examples: 636
      - name: en_US_Wavenet_A_1.0_0.0_0.0
        num_bytes: 140044143
        num_examples: 636
      - name: en_US_Wavenet_C_1.0_0.0_0.0
        num_bytes: 136477551
        num_examples: 636
      - name: en_US_Wavenet_A_1.5_0.0_0.0
        num_bytes: 94753263
        num_examples: 636
      - name: en_US_Wavenet_A_2.0_0.0_0.0
        num_bytes: 72080751
        num_examples: 636
      - name: en_US_Wavenet_A_0.5_0.0_0.0
        num_bytes: 276096495
        num_examples: 636
    download_size: 1401400700
    dataset_size: 1505649093
  - config_name: commoneval
    features:
      - name: audio
        dtype: audio
      - name: prompt
        dtype: string
    splits:
      - name: test
        num_bytes: 30944908
        num_examples: 200
    download_size: 27573315
    dataset_size: 30944908
  - config_name: ifeval
    features:
      - name: audio
        dtype: audio
      - name: key
        dtype: int64
      - name: prompt
        dtype: string
      - name: instruction_id_list
        sequence: string
      - name: kwargs
        list:
          - name: num_placeholders
            dtype: int64
          - name: capital_relation
            dtype: string
          - name: capital_frequency
            dtype: int64
          - name: keywords
            sequence: string
          - name: relation
            dtype: string
          - name: num_words
            dtype: int64
          - name: let_relation
            dtype: string
          - name: letter
            dtype: string
          - name: let_frequency
            dtype: int64
          - name: section_spliter
            dtype: string
          - name: num_sections
            dtype: int64
          - name: end_phrase
            dtype: string
          - name: forbidden_words
            sequence: string
          - name: keyword
            dtype: string
          - name: frequency
            dtype: int64
          - name: postscript_marker
            dtype: string
          - name: num_highlights
            dtype: int64
          - name: num_paragraphs
            dtype: int64
          - name: num_sentences
            dtype: int64
          - name: prompt_to_repeat
            dtype: string
          - name: first_word
            dtype: string
          - name: nth_paragraph
            dtype: int64
          - name: num_bullets
            dtype: int64
    splits:
      - name: test
        num_bytes: 126604577
        num_examples: 345
    download_size: 117898801
    dataset_size: 126604577
  - config_name: mmsu
    features:
      - name: audio
        dtype: audio
      - name: prompt
        dtype: string
      - name: reference
        dtype: string
      - name: question_id
        dtype: int64
      - name: cot_content
        dtype: string
      - name: category
        dtype: string
      - name: src
        dtype: string
    splits:
      - name: law
        num_bytes: 41103118
        num_examples: 51
      - name: engineering
        num_bytes: 79990255
        num_examples: 107
      - name: other
        num_bytes: 377556247
        num_examples: 546
      - name: biology
        num_bytes: 134336747
        num_examples: 172
      - name: business
        num_bytes: 186833430
        num_examples: 236
      - name: economics
        num_bytes: 229510852
        num_examples: 280
      - name: health
        num_bytes: 296546112
        num_examples: 406
      - name: philosophy
        num_bytes: 221002339
        num_examples: 305
      - name: psychology
        num_bytes: 248876187
        num_examples: 317
      - name: history
        num_bytes: 77305713
        num_examples: 104
      - name: chemistry
        num_bytes: 131897832
        num_examples: 167
      - name: physics
        num_bytes: 299038331
        num_examples: 383
    download_size: 2140740285
    dataset_size: 2323997163
  - config_name: mtbench
    features:
      - name: audio1
        dtype: audio
      - name: audio2
        dtype: audio
      - name: question_id
        dtype: int64
      - name: category
        dtype: string
      - name: turns
        sequence: string
      - name: reference
        sequence: string
    splits:
      - name: test
        num_bytes: 27206170
        num_examples: 46
    download_size: 25429481
    dataset_size: 27206170
  - config_name: openbookqa
    features:
      - name: audio
        dtype: audio
      - name: prompt
        dtype: string
      - name: reference
        dtype: string
    splits:
      - name: test
        num_bytes: 275196635
        num_examples: 455
    download_size: 251099084
    dataset_size: 275196635
  - config_name: sd-qa
    features:
      - name: audio
        dtype: audio
      - name: prompt
        dtype: string
      - name: reference
        dtype: string
    splits:
      - name: aus
        num_bytes: 75611830
        num_examples: 553
      - name: gbr
        num_bytes: 81039880
        num_examples: 553
      - name: ind_n
        num_bytes: 79954016
        num_examples: 553
      - name: ind_s
        num_bytes: 85840976
        num_examples: 553
      - name: irl
        num_bytes: 76020258
        num_examples: 553
      - name: kenya
        num_bytes: 58849868
        num_examples: 553
      - name: nga
        num_bytes: 98685976
        num_examples: 553
      - name: nzl
        num_bytes: 81047194
        num_examples: 553
      - name: phl
        num_bytes: 83175762
        num_examples: 553
      - name: usa
        num_bytes: 83810750
        num_examples: 553
      - name: zaf
        num_bytes: 75569934
        num_examples: 553
    download_size: 803403747
    dataset_size: 879606444
configs:
  - config_name: advbench
    data_files:
      - split: test
        path: advbench/test-*
  - config_name: alpacaeval
    data_files:
      - split: test
        path: alpacaeval/test-*
  - config_name: alpacaeval_full
    data_files:
      - split: test
        path: alpacaeval_full/test-*
  - config_name: alpacaeval_speaker
    data_files:
      - split: en_AU_Wavenet_A_1.0_0.0_0.0
        path: alpacaeval_speaker/en_AU_Wavenet_A_1.0_0.0_0.0-*
      - split: en_AU_Wavenet_B_1.0_0.0_0.0
        path: alpacaeval_speaker/en_AU_Wavenet_B_1.0_0.0_0.0-*
      - split: en_IN_Wavenet_A_1.0_0.0_0.0
        path: alpacaeval_speaker/en_IN_Wavenet_A_1.0_0.0_0.0-*
      - split: en_IN_Wavenet_B_1.0_0.0_0.0
        path: alpacaeval_speaker/en_IN_Wavenet_B_1.0_0.0_0.0-*
      - split: en_GB_Wavenet_A_1.0_0.0_0.0
        path: alpacaeval_speaker/en_GB_Wavenet_A_1.0_0.0_0.0-*
      - split: en_GB_Wavenet_B_1.0_0.0_0.0
        path: alpacaeval_speaker/en_GB_Wavenet_B_1.0_0.0_0.0-*
      - split: en_US_Wavenet_A_1.0_0.0_0.0
        path: alpacaeval_speaker/en_US_Wavenet_A_1.0_0.0_0.0-*
      - split: en_US_Wavenet_C_1.0_0.0_0.0
        path: alpacaeval_speaker/en_US_Wavenet_C_1.0_0.0_0.0-*
      - split: en_US_Wavenet_A_1.5_0.0_0.0
        path: alpacaeval_speaker/en_US_Wavenet_A_1.5_0.0_0.0-*
      - split: en_US_Wavenet_A_2.0_0.0_0.0
        path: alpacaeval_speaker/en_US_Wavenet_A_2.0_0.0_0.0-*
      - split: en_US_Wavenet_A_0.5_0.0_0.0
        path: alpacaeval_speaker/en_US_Wavenet_A_0.5_0.0_0.0-*
  - config_name: commoneval
    data_files:
      - split: test
        path: commoneval/test-*
  - config_name: ifeval
    data_files:
      - split: test
        path: ifeval/test-*
  - config_name: mmsu
    data_files:
      - split: law
        path: mmsu/law-*
      - split: engineering
        path: mmsu/engineering-*
      - split: other
        path: mmsu/other-*
      - split: biology
        path: mmsu/biology-*
      - split: business
        path: mmsu/business-*
      - split: economics
        path: mmsu/economics-*
      - split: health
        path: mmsu/health-*
      - split: philosophy
        path: mmsu/philosophy-*
      - split: psychology
        path: mmsu/psychology-*
      - split: history
        path: mmsu/history-*
      - split: chemistry
        path: mmsu/chemistry-*
      - split: physics
        path: mmsu/physics-*
  - config_name: mtbench
    data_files:
      - split: test
        path: mtbench/test-*
  - config_name: openbookqa
    data_files:
      - split: test
        path: openbookqa/test-*
  - config_name: sd-qa
    data_files:
      - split: aus
        path: sd-qa/aus-*
      - split: gbr
        path: sd-qa/gbr-*
      - split: ind_n
        path: sd-qa/ind_n-*
      - split: ind_s
        path: sd-qa/ind_s-*
      - split: irl
        path: sd-qa/irl-*
      - split: kenya
        path: sd-qa/kenya-*
      - split: nga
        path: sd-qa/nga-*
      - split: nzl
        path: sd-qa/nzl-*
      - split: phl
        path: sd-qa/phl-*
      - split: usa
        path: sd-qa/usa-*
      - split: zaf
        path: sd-qa/zaf-*
license: apache-2.0

License

The dataset is available under the Apache 2.0 license.

Citation

If you use the VoiceBench dataset in your research, please cite the following paper:

@article{chen2024voicebench,
  title={VoiceBench: Benchmarking LLM-Based Voice Assistants},
  author={Chen, Yiming and Yue, Xianghu and Zhang, Chen and Gao, Xiaoxue and Tan, Robby T. and Li, Haizhou},
  journal={arXiv preprint arXiv:2410.17196},
  year={2024}
}