AudioLLMs/gigaspeech2-test

这是一个包含多种低资源语言的大型自动语音识别（ASR）语料库，通过自动抓取、转录和精炼构建而成。语料库支持id-test、th-test和vi-test三种配置，每种配置都包含音频上下文、指令、答案以及额外的属性信息，如片段ID、开始时间、结束时间和原始文件路径。数据集适用于测试，并为每种配置提供了相应的数据文件路径。

This is a large-scale automatic speech recognition (ASR) corpus for multiple low-resource languages, constructed through automated crawling, transcription, and refinement. The corpus supports three configurations: id-test, th-test, and vi-test, each containing audio context, instruction, answer, and additional attribute information such as segment ID, start time, end time, and original file path. The dataset is suitable for testing and provides corresponding data file paths for each configuration.