ASCEND

该数据集是一个自发的、多轮对话式的语料库，特点是中文与英语之间的代码转换，该数据集在香港收集。此外，为了分析，该数据集被划分为三个子集：仅英语（EN）、仅普通话（ZH）以及混合集（包含普通话与英语代码转换的对话）。该数据集的任务是自动语音识别。

This dataset is a spontaneous multi-turn conversational corpus featuring code-switching between Mandarin and English, and it was collected in Hong Kong. For analytical purposes, the dataset is divided into three subsets: English-only (EN), Mandarin-only (ZH), and the mixed subset containing dialogues with Mandarin-English code-switching. This dataset is designed for the task of automatic speech recognition.