homebrew-research/instruction-speech-v1.5

该数据集包含多个特征，包括提示、回答、长度、索引、音频和标记。音频特征具有特定的采样率24000。数据集分为一个训练集，包含332,367个样本，总大小为120,758,849,451.125字节。下载大小为112,877,376,579字节。

The dataset contains multiple features including prompt, answer, length, index, audio, and tokens. The audio feature has a specific sampling rate of 24000. The dataset is divided into a training set with 332,367 samples, totaling 120,758,849,451.125 bytes in size. The download size is 112,877,376,579 bytes.