CMD

该数据集包含了大量的呼叫、短信和语音识别指令样本，这些样本短小精悍且包含了大量的实体信息。在数据处理方面，我们采用了内部的音频处理模型。在规模上，该数据集的训练样本有46万个，测试样本有1.4万个。所涉及的任务是自动语音识别（ASR）。

This dataset contains a large number of call, short message and speech recognition instruction samples, which are concise and contain abundant entity information. For data processing, an internal audio processing model was adopted. In terms of scale, the dataset has 460,000 training samples and 14,000 test samples. The target task is automatic speech recognition (ASR).