数据堂—41小时中国低幼儿童语音数据（麦克风+手机）

41小时中国低幼儿童语音数据由797位3至5岁中国儿童参与录制，其中5岁占比39%。录音内容符合儿童特点，主要为故事书、儿歌、口语等内容，每人120句，采用高保真麦克风与手机同步录制。有效数据时长41.8小时，文本经过人工转写，准确率高

This Chinese preschool children's speech dataset contains 41.8 hours of valid speech data, recorded by 797 Chinese children aged 3 to 5, with 39% of the participants being 5-year-olds. The recording content is tailored to children's age characteristics, mainly including storybook recitations, children's songs, spontaneous oral speech and other related contents. Each participant contributed 120 utterances. The recordings were collected synchronously using high-fidelity microphones and mobile phones. The corresponding transcripts were manually transcribed with high accuracy.

搜集汇总

背景与挑战

背景概述

该数据集包含41小时的中国低幼儿童语音数据，由797名3-5岁儿童使用麦克风和手机同步录制，内容涵盖故事书、儿歌等儿童适宜语音，并经过高精度人工转写。数据格式为16kHz/22.05kHz/44.1kHz的WAV文件，适用于儿童语音识别模型的测试任务。