数据堂— 592人数字类普通话方言手机采集语音数据

数字方言普通话手机采集语音数据，数据时长66小时；由592 名录音人参与录制，男女比例均匀；语种含四川方言、粤语、普通话；录制于日常生活场景；匹配主流安卓、苹果系统手机。

This is a speech dataset of digitized Mandarin dialects collected via mobile phones, with a total duration of 66 hours. A total of 592 speakers participated in the recording, with a balanced gender ratio. The dataset covers three speech varieties: Sichuan dialect, Cantonese and standard Mandarin. All recordings were conducted in daily life scenarios and were captured using mainstream Android and Apple iOS mobile phones.

搜集汇总

背景与挑战

背景概述

该数据集包含592名参与者录制的66小时语音数据，涵盖四川方言、粤语和普通话，用于中文语音识别模型的测试任务。数据格式为16kHz、16bit的单声道WAV文件，录制内容涉及日期、时间等数字相关句子，参与者中女性占比52%。