数据堂—300小时韩英混读手机采集语音数据

300小时韩英混读手机采集语音数据由韩国本土人员参与录制。录音文本均为韩英混合句子，涵盖通用场景及人机交互场景，内容丰富，转写精准。300小时韩英混读手机采集语音数据可用于改善语音识别系统对韩英混读语音的识别效果。

This 300-hour Korean-English code-switching speech dataset was collected via mobile phones, with recordings made by native Korean participants. All recorded speeches consist of mixed Korean-English sentences, covering both general scenarios and human-computer interaction scenarios, featuring rich content and accurate transcriptions. This dataset can be utilized to enhance the performance of speech recognition systems when processing Korean-English code-switching speech.

搜集汇总

背景与挑战

背景概述

该数据集包含300小时的韩英混读语音数据，由韩国母语者通过手机采集，用于测试韩英混读语音识别模型。数据格式为16kHz、16bit WAV单声道，涉及约600名参与者，版权归数据堂所有。