vanmanhnew/dataset

GigaSpeech 2是一个包含泰语、印度尼西亚语和越南语的大规模自动语音识别数据集。原始数据集包含约30,000小时的自动转录语音，精炼数据集包含10,000小时的泰语数据，以及各6,000小时的印度尼西亚语和越南语数据。此外，每个语言还有10小时的开发和测试集，由专业人类标注员转录，具有挑战性和现实性。数据集的音频文件被重新采样为16 kHz单通道WAV格式，文本经过标准化处理。

GigaSpeech 2 is a large-scale automatic speech recognition dataset containing Thai, Indonesian, and Vietnamese languages. The raw dataset includes approximately 30,000 hours of automatically transcribed speech, while the refined dataset consists of 10,000 hours of Thai data and 6,000 hours each for Indonesian and Vietnamese. In addition, there are 10 hours of development and test sets per language, transcribed by professional human annotators, which are challenging and realistic. The audio files in the dataset are resampled to 16 kHz single-channel WAV format, and the text has undergone standardization processing.