maxseats/aihub-464-preprocessed-680GB-set-51

该数据集包含音频、标签和输入特征三个主要特征。音频的采样率为16000Hz，标签为字符串类型，输入特征为浮点数序列。数据集分为训练集、测试集和验证集，分别包含45876、5735和5734个样本。下载大小为21994326235字节，数据集总大小为65723167273.99999字节。

This dataset is primarily used for audio processing tasks, including three main features: audio, labels, and input features. The audio feature has a sampling rate of 16000Hz, the label feature is of string type, and the input feature is a sequence of float32. The dataset is divided into training, testing, and validation parts, containing 45876, 5735, and 5734 samples respectively. The total download size of the dataset is 21994326235 bytes, and the total dataset size is 65723167273.99999 bytes. The configuration file defines the paths of the data files, corresponding to the training, testing, and validation sets.

原始信息汇总

数据集概述

数据特征

• 音频
  • 采样率: 16000
• 标签
  • 数据类型: 字符串
• 输入特征
  • 序列类型: 浮点数 (float32)

数据集划分

• 训练集
  • 字节数: 52578533819.2
  • 样本数: 45876
• 测试集
  • 字节数: 6572889777.947336
  • 样本数: 5735
• 验证集
  • 字节数: 6571743676.852664
  • 样本数: 5734

数据集大小

• 下载大小: 21994326235 字节
• 总大小: 65723167273.99999 字节

配置

• 默认配置
  • 训练集路径: data/train-*
  • 测试集路径: data/test-*
  • 验证集路径: data/valid-*