ZihanZhao/LibriSQA

---
task_categories:
- question-answering
language:
- en
size_categories:
- 100K<n<1M
configs:
- config_name: default
data_files:
- split: partI.train
path: "LibriSQA-PartI/LibriSQA-PartI-train.json"
- split: partI.test
path: "LibriSQA-PartI/LibriSQA-PartI-test.json"
- split: partII.train
path: "LibriSQA-PartII/LibriSQA-PartII-train.json"
- split: partII.test
path: "LibriSQA-PartII/LibriSQA-PartII-test.json"
---

# LibriSQA Dataset

- [LibriSQA Dataset](#librisqa-dataset)
- [Dataset Structure](#dataset-structure)
- [Keyword Explanation](#keyword-explanation)

## Dataset Structure

- `LibriSQA-PartI\LibriSQA-PartI-train.json`: metafile of train set of LibriSQA Part I
- `LibriSQA-PartI\LibriSQA-PartI-test.json`: metafile of test set of LibriSQA Part I
- `LibriSQA-PartII\LibriSQA-PartII-train.json`: metafile of train set of LibriSQA Part II
- `LibriSQA-PartII\LibriSQA-PartII-test.json`: metafile of test set of LibriSQA Part II

## Keyword Explanation

Explanation to each key in part I:
- speech_path: path to the speech
- question: question corresponding to the speech
- answer: reference answer regarding the speech
- text: authentic text corresponding to the speech
- duration: the duration of the speech

Explanation to each key in part II:
- speech_path: path to the speech
- question_and_option: question along with four options corresponding to the speech
- answer: correct answer label for the option (e.g. A)
- answer_and_analysis: reference answer and analysis regarding the speech
- text: authentic text corresponding to the speech
- duration: the duration of the speech

原始信息汇总

LibriSQA 数据集

数据集结构

• LibriSQA-PartI/LibriSQA-PartI-train.json: LibriSQA Part I 训练集的元文件
• LibriSQA-PartI/LibriSQA-PartI-test.json: LibriSQA Part I 测试集的元文件
• LibriSQA-PartII/LibriSQA-PartII-train.json: LibriSQA Part II 训练集的元文件
• LibriSQA-PartII/LibriSQA-PartII-test.json: LibriSQA Part II 测试集的元文件

关键词解释

Part I 关键词解释

• speech_path: 语音文件路径
• question: 与语音对应的问句
• answer: 与语音相关的参考答案
• text: 与语音对应的原始文本
• duration: 语音的时长

Part II 关键词解释

• speech_path: 语音文件路径
• question_and_option: 与语音对应的问句及四个选项
• answer: 选项的正确答案标签（例如 A）
• answer_and_analysis: 与语音相关的参考答案及分析
• text: 与语音对应的原始文本
• duration: 语音的时长