数据堂—1,652小时粤语手机采集语音数据

1,652小时粤语手机采集语音数据包括4888名来自广东省发音人，在安静的室内环境下的录音数据。录音内容广泛，覆盖50 万句常用口语语句，包括微博高频词、日常用语等。1,652小时粤语手机采集语音数据的句子平均重复次数1.5次，平均句长12.5字。匹配主流安卓、苹果系统手机

This dataset contains 1,652 hours of Cantonese speech data collected via mobile phones, with recordings made by 4,888 speakers from Guangdong Province in quiet indoor environments. The recorded content is extensive, covering 500,000 commonly used spoken sentences including high-frequency microblog terms and daily conversational expressions. For this dataset, the average repetition count per sentence is 1.5 times, and the average sentence length is 12.5 Chinese characters. It is compatible with mainstream Android and Apple smartphones.

搜集汇总

背景与挑战

背景概述

该数据集包含1,652小时的粤语手机采集语音数据，用于中文语音识别模型测试。数据由4,888名说话人在安静室内环境下录制，覆盖500,000个常用口语句子，采用16kHz WAV格式。