sartifyllc/Sartify_ITU_Zindi_Testdataset

这是一个为自动语音识别（ASR）和转录任务设计的音频文件数据集，作为Sartify比赛的一部分。数据集包含斯瓦希里语的音频样本，每个样本包括一个音频文件和其原始文件名，适合用于转录模型的测试和评估。数据集的结构包括音频的字节格式和文件路径信息，以及音频的元数据（如采样率）。目前数据集包含一个测试集，共有4089个样本，总时长约为7小时。

This dataset consists of audio files designed for automatic speech recognition (ASR) and transcription tasks as part of the Sartify competition. It includes Swahili language audio samples, with each sample comprising an audio file and its original filename, suitable for testing and evaluating transcription models. The dataset structure includes audio in byte format with file path information and audio metadata (such as sampling rate). Currently, the dataset contains a test split with a total of 4089 samples, spanning approximately 7 hours.