ARTPARK-IISc/Vaani-transcription-part

该数据集包含两个配置项：audio/Kurukh和audio/Tulu。每个配置项包含音频、说话者ID、语言、性别、州、地区、转录文本和参考图像等特征。数据集分为训练集、验证集和测试集，并提供了每个分割的字节数和样本数。audio/Kurukh配置项的训练集包含48个样本，验证集包含1个样本，测试集包含2个样本。audio/Tulu配置项的训练集包含909个样本，验证集包含67个样本，测试集包含116个样本。

The dataset includes two configurations: audio/Kurukh and audio/Tulu. Each configuration contains features such as audio, speakerID, language, gender, state, district, transcript, and referenceImage. The dataset is divided into train, validation, and test splits, with the number of bytes and examples provided for each split. The audio/Kurukh configuration has 48 examples in the train set, 1 example in the validation set, and 2 examples in the test set. The audio/Tulu configuration has 909 examples in the train set, 67 examples in the validation set, and 116 examples in the test set.