freococo/voa_myanmar_asr_audio_1

这是一个包含超过100万个音频块的自监督ASR-ready缅甸语音数据集，总时长约为3267小时。数据集来自Voice of America Burmese的广播节目，并经过分割处理。每个音频块都配有一个包含文件名、原始文件名、广播日期、链接和时长的.json文件。数据集适用于语音技术领域的训练和评估任务，尤其适合资源匮乏和低资源语言建模。数据集在语音建模和自监督学习方面具有高价值。

This is an ASR-ready Burmese speech dataset with over 1 million audio chunks, totaling around 3,267 hours of segmented audio. The dataset is derived from Voice of America Burmese radio broadcasts and has been segmented into chunks. Each audio chunk is paired with a .json file containing metadata such as file name, original file name, broadcast date, URL, and duration. The dataset is suitable for various speech technology tasks, particularly for under-resourced and low-resource language modeling. It is highly valuable for self-supervised learning and speech modeling.