ATCO2-PL-set corpus

该数据集包含了5281小时的未标记空中交通管制（ATC）数据，这些数据通过领域内语音识别器生成的自动转录文本进行了丰富，同时还加入了上下文信息、说话者轮转信息、每样本的信噪比估计以及英语语言检测分数。此外，数据集还增添了自动转录文本和其他补充的上下文信息。总体规模达到了5281小时的语音数据，其任务目标是自动语音识别。

This dataset contains 5,281 hours of unlabeled air traffic control (ATC) data, which is augmented with automatic transcriptions generated by in-domain speech recognizers, as well as contextual information, speaker turn information, per-sample signal-to-noise ratio (SNR) estimates, and English language detection scores. Additionally, the dataset also includes automatic transcriptions and other supplementary contextual information. The total scale of the dataset reaches 5,281 hours of speech data, and its target task is automatic speech recognition (ASR).