saurabh1043/noisy_kannada

该数据集是一个音频-文本对数据集，包含id、文本和音频三个特征字段。数据分为训练集（train）和开发集（dev），其中训练集有592,047个样本，总大小约163GB；开发集有31,160个样本，总大小约9.18GB。整个数据集下载大小约为171GB，总数据集大小约为172GB。This dataset is an audio-text pair dataset containing three feature fields: id, text, and audio. The data is divided into a training set (train) and a development set (dev), with the training set consisting of 592,047 samples and a total size of approximately 163GB, and the development set consisting of 31,160 samples and a total size of approximately 9.18GB. The entire dataset has a download size of about 171GB and a total dataset size of about 172GB.